Fracture Mechanics of Concrete and Concrete Structures -

Assessment, Durability, Monitoring and Retrofitting of Concrete Structures- B. H. Oh, et al. (eds)
ⓒ 2010 Korea Concrete Institute, Seoul, ISBN 978-89-5708-181-5

Numerical modeling of cracking of concrete due to corrosion of

reinforcement – Impact of cover thickness and concrete toughness
A.O.S. Solgaard
COWI A/S, Kgs. Lyngby, Denmark and Department of Civil Engineering, Technical University of Denmark
(DTU),Kgs. Lyngby, Denmark
A. Michel, H. Stang, & M. R. Geiker
Department of Civil Engineering, Technical University of Denmark (DTU),Kgs. Lyngby, Denmark
C. Edvardsen & A. Küter
COWI A/S, Kgs. Lyngby, Denmark

ABSTRACT: Deterioration of reinforced concrete structural members is an increasing problem in the con-
struction sector. The formation of corrosion products from steel reinforcing bars (rebars) leads to an increase
in volume and the possibility of cracking and spalling of the concrete as the tensile capacity of the concrete
matrix is exceeded. The ongoing work concerns a parametric study of cracking of cover concrete due to rein-
forcement corrosion. The basis of the simulations is a numerical fracture mechanical model for crack initia-
tion and propagation in concrete covers. The fracture mechanical model is based on the theory of the fictitious
crack, where the expansion of the corrosion products is modeled by the use of a fictitious thermal load. The
aim of the paper is to investigate the influence of the cover thickness and the ductility of the concrete on the
crack initiation and crack propagation due to reinforcement corrosion. The paper contributes to the under-
standing of cover cracking due to reinforcement corrosion with numerical investigations of the mechanical
parameters influencing the crack initiation and propagation.

1 INTRODUCTION and analytical models based on the theory of a thick-

Corrosion of reinforcement embedded in concrete is
one of the main deterioration processes of reinforced
concrete structures and has been the pivot of numer-
ous projects. As the corrosion products take up more
volume than the original steel consumed, a pressure
is build up in the interface between reinforcement
and concrete. The increase in pressure eventually
leads to cracking of the concrete cover. This allows
for faster ingress of de-passivating substances, e.g.
Cl ions and CO2, leading to faster deterioration of
the reinforcement.
Cracking of concrete due to reinforcement corro-
sion has been investigated experimentally and ana-
lytically in a number of papers. The experimental
work mainly concerns accelerated tests where a cur-
rent is applied to the reinforcement to control the
rate of corrosion e.g. Andrade et al. (1993) and
Alonso et al. (1998). The models for corrosion in-
duced cracking of concrete concerns the time to cor-
rosion induced cracking e.g. Liu & Weyers (1998),
modeling of cracking assuming smeared cracking,
e.g. Molina et al. (1993), modeling of cracking as-
suming an inner softening band, Noghaibi (1998)
walled cylinder, Chernin et al. (2009). In parts of the
literature experimental results are used for verifica-
tion of the presented models. However, there seems
to be an in-consistency between proposed model and
experimental data. This has resulted in the assump-
tion of a porous zone around the reinforcement,
where the corrosion products diffuse into without in-
creasing the pressure in the interface.
The scope of the present paper is to analyze crack
initiation and propagation in concrete due to rein-
forcement corrosion by the use of a numerical frac-
ture mechanical model based on the effective cohe-
sive relationship of the concrete. The relationship
between the thickness of the concrete cover and the
diameter of the reinforcement and crack initiation
and propagation is investigated for increasing thick-
nesses of corrosion products. Furthermore the initia-
tion and extent of crack propagation is modeled for
plain concrete (PC) and fibre reinforced concrete
(FRC) in order to investigate the impact of the in-
creased ductility of FRC. The latter is of great inter-
est, as small crack openings do not seem to have an
effect on the ingress of ions and CO2 leading to in-
creased corrosion of embedded reinforcement.
 As mentioned earlier cracking of cover concrete J = − D ( h, Tof)∇the
where tcor is the thickness h corrosion products
has been investigated in numerous research projects, and ηlin is the linear expansion coefficient of the cor-
experimentally as well as analytically. However, rosion products. TheThe linearproportionality coefficient
expansion coefficient for D(h,T)
simulations concerning the influence of the ductility moisture permeability and it is a nonlinea
an isotropic material is 1/3 of the volume expansion
of the concrete cover due to e.g. addition of fibres of thethat
coefficient. It is seen relative humidityofh the
the thickness andcorro-
temperature
have, to the knowledge of the author, not been inves- sion products tcor &canNajjar
be linked 1972). The moisture mass balanc
to time via Equations
tigated thoroughly. Hence the present paper contrib- 2-3. that the variation in time of the water mas
utes to these investigations with a modeling ap- volume of concrete (water content w) be eq
Description of model of the moisture flux J
divergence
proach for cracking of covers due to reinforcement
2.2
corrosion.
The expansion of the corrosion products is modeled
− ∂w = ∇thermal
by the use of a fictitious •J load. The approach
∂t
2 THEORY has earlier been used by e.g. Molina et al. (1993).
The modeling approach used in
The water the present
content w can context
be expressed a
is described in theoffollowing.
the evaporable water we (capillary wa
2.1 Corrosion of reinforcement 1) The rebar vapor,
with initial radius is water)
and adsorbed modeled,and cf.
the non-e
Corrosion of reinforcement embedded in concrete is Figure 1 for t=t0. (chemically bound) water w (Mil
n
a process influenced by numerous independent fac- 2) A prescribedPantazopoulo
thickness of consumed steel, X(t),
& Mills 1995). It is reas
tors. Depending on the environmental conditions, is applied to the assume
model, cf. thatFigure 1 for t=t1,1water
the evaporable . The is a fu
different corrosion processes can take place. Analy- original rebar is now transformed into two domains
relative humidity, h, degree of hydration
ses of concrete cover cracking due to corrosion of with different material
degreeproperties:
of silica fume An reaction,
inner core αs,ofi.e. we=w
reinforcement under laboratory conditions have been steel with Young’s modulus E and Poisson’s
= age-dependent sorption/desorption
st ratio
performed in numerous research projects, e.g. Val et γst and an outer perimeter
(Norling Mjonell 1997). Underwith
of corrosion products this assum
al. (2009), Isgor & Razaqpur (2006) and Caré et al. mechanical properties Young’s modulus E
by substituting Equation 1 into Equaticor and
(2008). However, there is a discrepancy between the Poisson’s ratio γcor .
obtains
different research projects which corrosion products 3) A thermal load, Tcor, corresponding to the lin-
to take into account. In the present context only the ear expansion coefficient of corrosion products is
∂w ∂h ∂w ∂w
formation of Fe2O3·H2O is taken into account. For- − e perimeter
applied to the outer + ∇ • ( D of
∇h)the e α& +The
= rebar. e α& + w
mation of other products with other expansion coef- h t h
thermal load of the outer perimeter causes
∂ ∂ ∂α c
α
an in-s s
∂
ficients is beyond the scope of the present work. The crease in the volume of the corrosion products, cf.
c
corrosion is assumed uniform, and a sketch of the re- Figure 1 for t=t1,2.
inforcement is given in Figure 1.
where ∂we/∂h is the slope of the sorption/
Diffusion of corrosion products into the concrete
isotherm (also called moisture capac
matrix is not taken into account in the present
governing equation (Equation 3) must be
model. Hence the volume of corrosion products
by appropriate boundary and initial conditi
leading to an increase in the pressure in the interface
The relation between the amount of e
is proportional to the volume of consumed steel.
water and relative humidity is called ‘‘
isotherm” if measured with increasing
humidity and ‘‘desorption isotherm” in th
(1) case. Neglecting their difference (Xi et al.
the following, ‘‘sorption isotherm” will be
where ∆Vc is the volume of corrosion products, ∆Vs reference to both sorption and desorption c
is the volume of consumed steel and ηvol is the vol- Figure 1. Schematic By the way,
illustration if thethehysteresis
of modeling expansion ofof the
corrosion products.
ume expansion coefficient of the corrosion products. isotherm would be taken into account, two
Assuming uniform corrosion, the depth of con- Please note that relation,
t1>t0 andevaporable water vs relative
that the modeling is di- humi
sumed steel due to corrosion can be determined from be used according to the sign
vided in two steps for t=t1 viz. t1,1 and t1,2. Further- of the varia
Faraday’s law, cf. Equation 2: more, the free expansion of the corrosion products, of the
relativity humidity. The shape
isotherm
as shown to the right for HPC
of Figure 1 doesis influenced by many p
not take place
especially those that
when the rebar is embedded in the concrete body. influence extent and
(2) chemical reactions and, in turn, determ
where M is the atomic mass of iron, ρ is the density As the thickness of the corroded steel, X(t), has to
be prescribed prior structure and pore
to modeling it issize
notdistribution
possible to (water-
of iron, n is the valence of iron, F is Faraday’s number, ratio, cementmodeling
chemical composition, SF
t is the time and icor is the corrosion current density. use the model for transient of the crack
curing time and method,
propagation due to an increase in corrosion products. temperature, mix
As the thickness of the corrosion products is very
small compared to the radius of the reinforcement, Thus in order toetc.). In the
obtain the literature various and
crack initiation formulatio
we get: propagation over time, meaning for an increase isotherm
found to describe the sorption in
X(t), cf. Equationconcrete (Xi et al.to1994).
2 it is necessary However,
run a series of in th
paperX(t).
analyses for different the semi-empirical expression pro
(3) Norling Mjornell (1997) is adopted b

Proceedings of FraMCoS-7, May 23-28, 2010

J − D (variation
=The h, T )∇h in mechanical properties and tem-
(1) explicitly accounts
The elements for the
outside evolution
the crack of hydration
are considered to
perature through the cross-section for t=t1 is reactionlinear
behave and elastically
SF content.andThis sorption isotherm
isotropically, whereas
The proportionality
sketched in Figure 2: coefficient D(h,T) is called reads
the elements in the interface are behaving non-linear
moisture permeability and it is a nonlinear function according to a predefined cohesive law giving the
of the relative humidity h and temperature T (Bažant relationship between stress ⎡ and crack opening. ⎤ A to-
& Najjar 1972). The moisture mass balance requires tal of app. 900 elements
we (h, α c ,have ⎢ each with
α s ) = G (α c , α s )⎢1 −
1 16 degrees
⎥
+ of
that the variation in time of the water mass per unit freedom been
1 used for the mesh.
10(g α
∞ ⎥
− α )h ⎥ ini-
volume of concrete (water content w) be equal to the In the modeling approach ⎢
⎣ e used1 cfor thec crack
⎦ (4)
divergence of the moisture flux J tiation and propagation it is assumed that:
⎡ 10(g α ∞ − α c )h ⎤
K1 (α c , α s )⎢e 1 c − 1⎥
− ∂ = ∇•J
w
(2) ⎢
⎣
⎥
⎦ (4)
∂ t

The water content w can be expressed as the sum where σthe

where first term (gel isotherm) represents the
e is the stress in the elastic domain, Ec is
of the2.evaporable (capillary
water wofe the water, water physically
Youngs modulus forbound (adsorbed)
concrete,waterε is the andelastic
the second
strain,
Figure
vapor,
cal
Schematic illustration
and adsorbed
properties water)across
and temperature
distribution
and rebar
of mechani-
the non-evaporable
cross section for
term (capillary isotherm) represents
σw is the bridging stress across the crack, g(w) the capillary
is the
(chemically
t=t1. bound) water wn (Mills 1966, water. This expression
dimensionless softeningisfunction
valid only and for
ft islow
the content
tensile
Pantazopoulo & Mills 1995). It is reasonable to of SF. The
strength coefficientThe
of concrete. represents
G1cohesive laws theused
amount in theof
assume
Pleasethat
notethe
thatevaporable
in Figure 2, water is a function
the variations of
of me- water per unit volume held in the
present context are described in detail in the modelgel pores at 100%
relative humidity, h, degree of hydration, αc, and
chanical properties are not in scale. The thermal load relative humidity, and it can be expressed (Norling
description.
degree of
applied to silica fume reaction,
the perimeter αs, i.e.Tcor
of the rebar =we(h,αc,αs)
we=2.1. Mjornell 1997) aslayer is placed right above the rein-
The interface
= age-dependent sorption/desorption isotherm
The interaction between the reinforcement and forcement through the cover concrete. Experimental
(Norling
the Mjonell
confining 1997).
concrete Under inthis
is shown assumption
Figure 3. and investigations c ofα cracking
G1 (α c , α s ) = k vg s α sof concrete due to rein-
c + k vg (5)
by substituting Equation 1 into Equation 2 one forcement corrosion, e.g.s Liu & Weyers (1998),
c
obtains have showed that cracks are mainly formed verti-
whereabove
cally kcvg andtheksvgreinforcement
are material parameters.
through the From the
cover
∂w ∂h maximum
layer. Thusamount of waterofpertheunitpre-defined
the position volume thatcrack can
− e + ∇ • ( D ∇h) = ∂we α ∂we α&+ & + w& (3) fill allthe
path, pores (both capillary
interface layer, is pores and gel pores),
in accordance with one ex-
∂h ∂t h ∂α c
∂α s n
perimental K1 as one obtains
can calculateobservations. The mesh used for the
c s
Figure 3. Schematic illustration of pressure on confining con- modeling is shown in Figure 5:
crete due to expansive corrosion products. Adapted from Liu &
Weyers ∂we/∂h is the slope of the sorption/desorption
where (1998). ⎡
⎢ ⎜ g α − α ⎟h ⎥
⎛
10
c
∞
c
⎞ ⎤

isotherm (also called moisture capacity). The w − 0.188 α s + α s −G ⎢ −e

c 0.22
s 1
⎝ 1 ⎠
⎥

governing equation
3 it (Equation
is seen that3)the
must be completed ⎦ (6)
0 1
⎢ ⎥
From Figure increase of vol- K (α c α s ) = ⎣
by appropriate boundary and initial conditions.
ume due to corrosion causes pressure in the steel- 1
,

⎜ g α − α ⎟h
⎛ ∞ ⎞

The relation
interfacebetween
and thisthe amount of beevaporable e ⎝ c c⎠ −
10

concrete pressure must counter- 1

1

waterbyand
acted therelative
confining humidity
concrete.is called ‘‘adsorption
isotherm” if measured with increasing relativity The material parameters kcvg and ksvg and g1 can
humidity and ‘‘desorption isotherm” in the opposite be calibrated by fitting experimental data relevant to
2.3
case.Fracture mechanics
Neglecting of concrete
their difference (Xi et al. 1994), in free (evaporable) water content in concrete at
the following,
The ‘‘sorption
crack initiation andisotherm” will bein used
propagation with
a semi- various ages (Di Luzio & Cusatis 2009b).
referenceconcrete
infinite to both sorption
body has andbeen
desorption
modeled conditions.
in the Figure 5. Mesh for modeling the cracking of cover due to rein-
By the way,FE-program
commercial if the hysteresis
DIANA by of the
the usemoisture
of a forcement corrosion.
2.2 Temperature evolution
isotherm crack
discrete wouldformulation
be taken intobased
account, two fictitious
on the different
relation,
crack evaporable
model suggestedwaterbyvsHillerborg
relative humidity, must
et al. (1976). Note
Thethat,
boldat lines
earlyindicate
age, since
thethe chemical
position reactions
of the inter-
be used
The crossaccording
section oftothe
themodel
sign ofused
theisvariation
shown inofFig-the associated with cement hydration and SF reaction
face layer. It is seen from Figure 5, that the meshing
relativity
ure 4. humidity. The shape of the sorption aredenser
is exothermic, the temperature
at the interface fieldmeshing
layer. The is not uniform
for the
isotherm for HPC is influenced by many parameters, for non-adiabatic systems even if the environmental
reinforcement and the interface layer is shown in
especially those that influence extent and rate of the temperature is constant. Heat conduction can be
Figure 6:
chemical reactions and, in turn, determine pore described in concrete, at least for temperature not
structure and pore size distribution (water-to-cement exceeding 100°C (Bažant & Kaplan 1996), by
ratio, cement chemical composition, SF content, Fourier’s law, which reads
curing time and method, temperature, mix additives,
etc.). In the literature various formulations can be q = − λ ∇T (7)
found to describe the sorption isotherm of normal
concrete (Xi et al. 1994). However, in the present where q is the heat flux, T is the absolute
paper the semi-empirical expression proposed by temperature, and λ is the heat conductivity; in this
Norling4. Cross
Figure Mjornell
section(1997)
of modelisusedadopted because ofit
for determination
crack initiation and crack propagation. Figure 6. Meshing of reinforcement and interface layer.

Proceedings of FraMCoS-7, May 23-28, 2010

 The measures of the semi-infinite concrete body = − D (h, T )to
cantly. However, Jaccording ∇hLöfgren et al. (2008),
are 100 mm on each side of the reinforcement, ex- the addition of fibres to the matrix changes the
cept from cover, which has revealed to be sufficient stress-crack openingThe proportionality coefficient D(h,T)
relationship.
to ensure that the behavior of the body is semi- moisture permeability
According to Hillerborg andtheit tough-
et al. (1976), is a nonlinea
infinite. Modeling is conducted for plain strain and of the relative
ness of concrete should humidity
be determined h and
from the temperature
frac-
cracking is solely modeled as Mode-I, opening ture energy, the&elastic
Najjar 1972). The moisture mass balanc
modulus and the tensile
mode, cracking. strength: that the variation in time of the water mas
A slip interface has been modeled around the up- volume of concrete (water content w) be eq
per half of the rebar. The interface elements are used divergence of the moisture flux J
for simulation of the debonding in the steel-concrete (5)
interface and have Mixed-Mode properties and a
− ∂w = ∇ of
where lch is the characteristic length of the material,
shear retention factor of app. 3. Ec is the elastic modulus • J concrete, Gf is the frac-
∂t
Further descriptions of the model are given in ture energy and ft is the tensile strength. Note that
Michel et al. (2009). The water
the characteristic length contentproportional
is inverse w can be expressed
to a
of squared.
the tensile strength the evaporableThus high water we (capillary
strength ma- wa
terials, which arevapor, andbrittle
usually adsorbedwillwater)
get a and lowerthe non-e
3 MODELING OF CRACKING (chemically bound) water with
characteristic length than low strength materials wn (Mil
the same elasticPantazopoulo
modulus and & Mills
fracture 1995).
energy It is reas
The crack initiation and propagation in reinforced assume that the evaporable water is a fu
Rosello et al. (2005).
concrete due to reinforcement corrosion is modeled Two differentrelative humidity,
approaches , degree
for hthe mechanicalof hydration
for various cover thicknesses and various mechani- degree of silica fume reaction, αs, i.e. we=w
properties of the cover layer have been used:
cal properties of the cover layer, viz. cohesive rela- 1) The slope of= theage-dependent
second branch, asorption/desorption
2, is constant
tionships. (Norling Mjonell
and the intersection of the second1997).branch
Underwith this assum
by substituting Equation 1 into Equati
the ordinate-axis, b 2 , is shifted in equal inter-
3.1 Parameters for the model vals. The obtains
cohesive parameters are given in
Table 2.
In order to describe the influence of the cover layer 2) The slope of∂thewe ∂second branch, a∂2w, is halved
h + ∇approaches e α&the+ ∂in-we
on the cracking, different values for the thickness of between the − different • ( D ∇h) = and
h with α&s + w
the cover layer have been investigated, viz. 10, 20 ∂h second
tersection of the ∂t branch ∂α the ordi-
c
∂α
and 40 mm. For all of the simulations, the diameter
c s
nate axis, b2, is shifted in equal intervals. The
of the reinforcement bar is constant. The values for cohesive parameters
the mechanical properties of the reinforcement, the
where ∂weare /∂h given
is theinslope
Table of3.the sorption/
In the following, isotherm
the cohesive relationships
(also called moisture modeled capac
elastic domain of concrete and corrosion products from approach 1governing
are namedequation b2,1-b2,4(Equation
, and the 3) rela-
are constant for all of the simulations and given in
must be
tionships obtainedbyfrom approach 2 are named
appropriate boundary and initial a2,1- conditi
Table 1: a2,4. The relation between the amount of e
Table 1. Constants for fracture mechanical model.
water and relative humidity is called ‘‘
Table 2. Material parameters for modeling approach 1.
Parameter Value Dimension isotherm”
b2,1 b2,2
if measured
b2,3
with
b2,4
increasing
humidity and ‘‘desorption isotherm” in th
Young’s modulus for steel 210 GPa a1 [mm-1] case. Neglecting
27.5 19.2 their difference
10.8 2.5 (Xi et al.
Young’s modulus for concrete 31 GPa the following, ‘‘sorption isotherm” will be
a2 [mm-1] reference2.5to both2.5
2.5 sorption-and desorption c
Young’s modulus for corr. product 2.1 GPa
Tensile strength of concrete 3.0 MPa b2 [-] By the0.67
0.5 way, if0.83the hysteresis
1 of the
isotherm would be taken into account, two
Poisson ratio of steel 0.3 - Gf [J/m2] relation,277
165 evaporable421 water600 vs relative humi
Poisson ratio of concrete 0.2 - lch [mm] be
568 used according
954 1450to the sign
2067 of the varia
Poisson ratio of corr. products 0.2 - relativity humidity. The shape of the
m/K-1
isotherm for HPC is influenced by many p
Expansion coef. of corr. product 2.1
especially
Table 3. Material parameters those that
for modeling influence
approach 2. extent and
Diameter of reinforcement 10 mm chemicala2,2 reactions
a2,1 a2,3 and, ain 2,4 turn, determ

a1 [mm-1]
structure19.2
27.5
and pore110size distribution
0.3
(water-
In order to investigate crack initiation and propa- ratio, cement chemical composition, SF
gation for different mechanical properties of con- a2 [mm-1] curing time
2.5 1.2 and method,
0.6 temperature,
- mix
crete the cohesive relationships for plain concrete b2 [-]
etc.).
0.5
In the literature various formulatio
and fibre reinforced concrete are investigated. found to0.67describe0.83the sorption
1
isotherm
For the modeling of the cohesive relationship, a
2
Gf [J/m ] concrete (Xi et al. 1994). However, in th
165 543 1685 4800
bi-linear model has been suggested. For plain con- lch [mm]
paper the
568
semi-empirical expression pro
crete the cohesive relationship does not vary signifi- Norling1870Mjornell5804(1997)16533 is adopted b

Proceedings of FraMCoS-7, May 23-28, 2010

 J − D (hthat
=Note , T ) ∇h
the cohesive relationships b2,1 and (1)
a2,1 explicitly
Once theaccounts
crack is for the evolution
initiated and starts of hydration
to propagate
equals as these are the relationships forming the ba- itreaction and that
is assumed SF the
content. Thisproducts
corrosion sorptiondoisotherm
not in-
sis The
of theproportionality coefficient
other relationships. These D(h,T) is called
relationships re- reads into the crack. Furthermore the porous ITZ
trude
moisture permeability and it is a nonlinear function
flect the mechanical properties of plain concrete. around the reinforcement is not taken into account,
of thecohesive
The relativerelationships
humidity h andaretemperature
illustrated inT (Bažant
Figures thus it is assumed that the ⎡ corrosion products ⎤ do not
& Najjar 1972). The moisture mass balance requires
7-8: diffuse into the ITZ but contributes
we (h, α c , in ⎢
α s ) = G (α c , α s )⎢1 −
to the increase
1 ⎥ in
+ and
that3the variation in time of the water mass per unit pressure the interface between ∞
reinforcement
10(g α
⎥
c )h ⎥⎦ (4)a
− αrepresents
1

1 c
volume of concrete (water content w) be equal tob the concrete. Thus the presented ⎢
⎣ e model
divergence of the moisture flux J conservative simulation of the buildup of pressure
2,1
b ⎡ 10(g α ∞ − α )h ⎤
2.5
2,2
due to corrosion of reinforcement. 1 c c
b2,3
Due to limitationsK1 (α c , αin ⎢ e FE-modeling− 1software
s )⎢the
⎥
− ∂ 2 = ∇•J w
(2)
b2,4
the minimum thickness of ⎣ the consumed steel,
⎥
⎦ X(t)
∂
Bridging stress [MPa]

t
cf. Figure 1 is 2.5µm. Consequently the crack initia-
The water content w can be expressed as the sum where
tion the first
cannot term (gelforisotherm)
be simulated represents
small cover the
thicknesses
1.5
of the evaporable water we (capillary water, water physically
with the presentboundmodel.
(adsorbed) water and the second
vapor, and adsorbed water) and the non-evaporable term (capillary isotherm) represents the capillary
(chemically bound) water wn (Mills 1966,
1 water. This expression is valid only for low content
3.3 Results
of SF. from modeling
The coefficient G1 represents the amount of
Pantazopoulo & Mills 1995). It is reasonable to watermodeling
per unit volume held in the gelforpores at 100%
assume
0.5
that the evaporable water is a function of The has been conducted small crack
relative humidity, h, degree of hydration, αc, and relative humidity,
openings and it Thus
at the surface. can betheexpressed
thickness(Norling
of the
degree
0 of 0.05
silica 0.1
fume 0.15
reaction, , i.e. w0.3e=we(0.35
αs0.25 s)
h,αc,α0.4 Mjornell 1997)
corrosion as is varied between app. 5µm and
products
= age-dependent sorption/desorption
0 0.2
isotherm 21µm. Thus it is not possible to investigate the crack
= k csmall s
CMOD [mm]
(Norling Mjonelllaws1997). Under this initiation
G1 (α c , α s )for kcover thicknesses (10mm and
Figure 7. Cohesive for approaches b2,1-bassumption
2,4.
and
20mm). vg α ctoc +investigate
In order vg α s s the crack propagation (5)
by substituting Equation 1 into Equation 2 one
obtains
3
through the concrete cover the crack opening at the
a2,1 steel kcvg and1.5
wheresurface, ksvgmm are above
materialtheparameters. From
steel surface andthe
at
w ∂h
∂2.5 ∂w ∂w
a2,2 maximum amount of water per unit volume that can
the concrete surface have been calculated for all ge-
− e + ∇ • ( D ∇h ) = e α&c + e α&s + w&n (3) fill all pores
ometries and (both
cohesive capillary pores and gel pores), one
relationships.
a2,3
∂h ∂t h ∂α ∂α can calculate K1 as one obtains
2
c s a2,4

3.3.1 Crack initiation and propagation

Bridging stress [MPa]

The crack initiation and propagation ⎛ at ∞different

⎞ ⎤ lo-
where ∂we/∂h is the slope of the sorption/desorption cations through the cross
⎡
section
⎢ ⎜ g α − α ⎟h ⎥
for c c
10mm and
10

isotherm
1.5
(also called moisture capacity). The w − α s + α s −G ⎢ −e
c
40mm cover thickness for plain s
0.188 0.22
⎝ ⎠1⎥
⎢ concrete, approach
1

governing equation (Equation 3) must be completed ⎦ (6)

0 1
⎥

bK2,1(α, care
α )shown in Figures 9-10: ⎣
by appropriate
1 boundary and initial conditions. s =
1
,

⎜ g α − α ⎟h
⎛ ∞ ⎞

The relation between the amount of evaporable e ⎝ c c⎠ −

10
1
0.09 1

water
0.5 and relative humidity is called ‘‘adsorption At steel
isotherm” if measured with increasing relativity The material parameters kcvg and ksvg and g1 can
0.08 1.5mm above steel

humidity and ‘‘desorption isotherm” in the opposite be 0.07

calibrated by fitting experimental data relevant to
At concrete surface

case. Neglecting their difference 2 (Xi 2.5et al. 1994), in

0
0 0.5 1 1.5 3 3.5 free (evaporable) water content in concrete at
the following, ‘‘sorption isotherm” will be used with
CMOD [mm]
various
0.06 ages (Di Luzio & Cusatis 2009b).
Crack opening [mm]

reference
Figure to bothlaws
8. Cohesive sorption and desorption
for approaches a2,1-a2,4. conditions.
By the way, if the hysteresis of the moisture 0.05
2.2 Temperature evolution
Thus in
isotherm total be
would seven
takendifferent cohesive
into account, two relation-
different 0.04
ships are investigated. Note the difference
relation, evaporable water vs relative humidity, in themust
or- Note that, at early age, since the chemical reactions
dinate axis for Figures 7 and 8. According
be used according to the sign of the variation of the to Kazemi associated
0.03 with cement hydration and SF reaction
et al. (2007),
relativity the fracture
humidity. Theenergy
shapeis increased app. 11
of the sorption are0.02exothermic, the temperature field is not uniform
times
isotherm whenforsmall
HPC amounts
is influenced(0.5 by vol-%-1.5 vol-%) of
many parameters, for non-adiabatic systems even if the environmental
32mm steel fibres are added to plain concrete.
especially those that influence extent and rate of the temperature
0.01 is constant. Heat conduction can be
chemical reactions and, in turn, determine pore described in concrete, at least for temperature not
structure
3.2 and pore
Limitations forsize distribution
modeling approach (water-to-cement exceeding
0.004 0.006100°C
0
(Bažant
0.008 0.01 0.012 &0.014 Kaplan 1996),
0.016 0.018 by
0.02 0.022
ratio, cement chemical composition, SF content, Fourier’s law, which reads t [mm] cor
The
curingcracktimeinitiation
and method, and temperature,
crack propagation in rein-
mix additives, Figure 9. Results from modeling of crack initiation and propa-
forced concrete described in the
etc.). In the literature various formulations can bepresent paper is q = −in
λ ∇plain
T concrete, b2,1, for C=10mm.
modeled by the use of a fracture mechanical model,
gation
(7)
found to describe the sorption isotherm of normal
where
concrete a fictitious
(Xi et al.thermal
1994).load is applied
However, in to
thethepresent
rein-
forcement in order to simulate the
paper the semi-empirical expression proposed by expansive corro- where q is the heat flux, T is the absolute
sion products.
Norling This modeling
Mjornell (1997) isapproach adoptedhasbecause
also beenit temperature, and λ is the heat conductivity; in this
used by e.g. Molina et al. (1993).
Proceedings of FraMCoS-7, May 23-28, 2010
 = − D (h,and
3.3.3 Analyzes ofJCMOD T )∇hcover thickness
0.08
At steel In order to evaluate the influence of the ductility of
0.07 1.5mm above steel
At concrete surface
the cover concrete on Thetheproportionality
crack initiation coefficient
and propa- D(h,T)
gation, the crack moisture
opening atpermeability
the concreteand it is and
surface a nonlinea
0.06
of the
at the concrete-steel relativehave
interface humidity and temperature
been hplotted for
& Najjar 1972). The moisture mass balanc
various cover thicknesses and various material pa-
Crack opening [mm]

0.05

that13-14.
rameters in Figures the variation in time of the water mas
volume of concrete (water content w) be eq
0.04

divergence of the moisture flux J

0.08
0.03
At steel a C=10 mm
2,4
0.07 At conrete surface a2,4 C=10 mm
0.02

− ∂ = ∇•
At steel b2,1 C=20
w mm
J
0.01 0.06
∂
At concrete surface
t b2,1 C=20 mm

Crack opening [mm]

0 0.05
0.004 0.006 0.008 0.01 0.012 0.014
tcor [mm]
0.016 0.018 0.02 0.022
The water content w can be expressed a
Figure 10. Results from modeling of crack initiation and propa-
of the evaporable water we (capillary wa
0.04

gation in plain concrete, b2,1, for C=40mm. vapor, and adsorbed water) and the non-e
0.03
(chemically bound) water wn (Mil
3.3.2 Influence of cohesive relationship on cracking Pantazopoulo & Mills 1995). It is reas
assume that the evaporable water is a fu
0.02
In order to illustrate the influence of the cohesive re-
lationship on the cracking of the cover layer, the 0.01 relative humidity, h, degree of hydration
displacements at different locations for the cohesive degree of silica fume reaction, αs, i.e. we=w
0.004 0.006 0.008= 0.01age-dependent
0.012 0.014 0.016 sorption/desorption
0
relationships for plain concrete, b2,1, and the most 0.018 0.02 0.022

ductile FRC investigated, a2,4, for C=10mm and (Norlingt Mjonell

[mm] 1997). Under this assum
cor

C=40mm are illustrated in Figures 11-12: by substituting

Figure 13. Crack opening at concrete Equation
surface and 1at into
steel- Equati
concrete interface forobtains
C=20mm and b2,1 vs. C=10mm and a2,4.
0.09
At steel b2,1
we h ∂w
e α& + ∂we α& + w
0.08
0.08 1.5mm above steel b2,1 At steel a2,4∂C=20 ∂
mm
− +∇•( D h ∇h ) =
∂h ∂t ∂α ∂α
At concrete surface b2,1 0.07 At concrete surface a2,4 C=20 mm c s
c s
0.07
At steel a2,4 At steel b2,1 C=40 mm
0.06 1.5mm above steel a2,4 0.06
At concrete surface b2,1 C=40 mm
Crack opening [mm]

0.05
At concrete surface a2,4 where ∂we/∂h is the slope of the sorption/
Crack opening [mm]

isotherm (also called moisture capac

0.05
0.04
0.04 governing equation (Equation 3) must be
0.03 by appropriate boundary and initial conditi
0.02
0.03 The relation between the amount of e
water and relative humidity is called ‘‘
0.01 0.02
isotherm” if measured with increasing
0
0.01
humidity and ‘‘desorption isotherm” in th
0.004 0.006 0.008 0.01 0.012 0.014
tcor [mm]
0.016 0.018 0.02 0.022
case. Neglecting their difference (Xi et al.
Figure 11. Results from modeling of crack initiation and propa- 0 the0.01
following, ‘‘sorption isotherm” will be
referencet to[mm]both sorption and desorption c
0.004 0.006 0.008 0.012 0.014 0.016 0.018 0.02 0.022
gation in plain concrete and FRC for C=10mm.
By theat concrete
way, ifsurfacethe hysteresis
and at steel-of the
cor

Figure 14. Crack opening

concrete interface forisotherm would beC=20mm
taken intoand account, two
0.08
At steel b2,1 C=40mm and b2,1 vs. a2,4.
0.07 1.5mm above steel b2,1 relation, evaporable water vs relative humi
At concrete surface b2,1
3.3.4 Analyzes ofbetheused twoaccording
approaches to for
themodeling
sign of the varia
relativity humidity. The shape of the
0.06
At steel a2,4
the cohesive relationship
1.5mm above steel a2,4
isotherm forinvestigated
HPC is influenced by many p
Crack opening [mm]

0.05
The cohesive relationships in the pre-
sent paper are especially those that
two influence
different extent
ap- and
At concrete surface a2,4
modeled from
proaches cf. Section 3.1. Thus the difference be- determ
chemical reactions and, in turn,
0.04

0.03
tween the two structure approaches and poreis size distribution
investigated by (water-
0.02 analyzing the crack ratio, cementat the
opening chemical
concrete composition,
surface SF
for cover thicknesses curing10mm
time andand 40mm.
method,The temperature,
results mix
0.01
are shown in Figures etc.).15-16:
In the literature various formulatio
0 found to describe the sorption isotherm
0.004 0.006 0.008 0.01 0.012 0.014
tcor [mm]
0.016 0.018 0.02 0.022
concrete (Xi et al. 1994). However, in th
Figure 12. Results from modeling of crack initiation and propa- paper the semi-empirical expression pro
gation in plain concrete and FRC for C=40mm. Norling Mjornell (1997) is adopted b

Proceedings of FraMCoS-7, May 23-28, 2010

 J = − D ( h , T ) ∇h (1) explicitly
opens accounts
linearly with for the the evolution
increase in the of corrosion
hydration
0.09
reaction and
thickness. SF content.
Comparing FigureThis 9 andsorption
10 it is seen,isotherm
that
The proportionality
b 0.08
b2,1
coefficient D(h,T) is called reads
an increase in the cover thickness, requires a larger
moisture permeability and it is a nonlinear function
2,2
b
corrosion thickness to initiate a crack. Additionally
of 0.07
the relative
b humidity h and temperature T (Bažant
2,3
it is seen from Figures 9-10 ⎡ that the crack propaga- ⎤
& 0.06
Najjar 1972). The moisture mass balance requires tion changes significantly when the cover⎥ is in-
2,4
⎢
a we (h, α c from
, α ) = G (α , α ) 1 −
s 10mm c tos 40mm.
1
+
∞
Crack opening [mm]

that the variation in time of the water mass per unit creased For the values ⎥ of tcor
2,2
⎢
10(g α − α c )h propaga-
1
a
volume of concrete (water content w) be equal to the simulated in the present ⎢⎣context, e 1thec crack ⎥
2,3

0.05 a ⎦ (4)
divergence of the moisture flux J
2,4
tion for C=40mm is stable inside the concrete cover
⎡ 10(g α ∞ − α )h ⎤
and propagates until it reaches the concrete
1 c c surface.
This is not theKcase
(α , α ) e
s ⎢C=10mm where− the
0.04 ⎢ 1⎥
c for ⎥ crack
− ∂0.03= ∇ • J (2)
1
w

∂ t
propagates to the surface ⎣for very small values ⎦ of tcor
(< 5µm).
0.02
The water content w can be expressed as the sum where the first
Regarding thetermeffect(gelof the isotherm)
mechanical represents
properties the
of 0.01
the evaporable water we (capillary water,0.02 water physically
of the concrete bound on (adsorbed) water andandthepropaga-
the crack initiation second
vapor, and adsorbed water)
0.004 0.006 0.008
and the non-evaporable
0.01 0.012 0.014 0.016 0.018 0.022 term it(capillary
tion, is seen from isotherm)
Figure 11 represents
that the crack the capillary
opening
at the steel-concrete interface is regardless of content
water. This expression is valid only for low
t [mm]
(chemically bound) at concrete
water surface
wn for (Mills
cor
1966, the co-
Figure 15. Crack opening
Pantazopoulo & Mills 1995). It is
C=10mm
reasonable
and
to of SF. relationship
hesive The coefficient G1 represents
for materials with thethe amount
same ten- of
various approaches for the cohesive relationship.
assume that the evaporable water is a function of waterstrength.
sile per unitConsidering
volume heldthe in the
crack gelopening
pores at at100% the
relative
0.08 humidity, h, degree of hydration, αc, and relative of
surface humidity,
the concrete,and it whichcan be isexpressed
a governing (Norlingpa-
degree of silica
b
fume reaction, αs, i.e. we=we(h,αc,αs)
2,1
Mjornell 1997) as
rameter when discussing cracking of concrete, it is
= 0.07age-dependent
b
sorption/desorption isotherm
2,2 seen from Figure 11, that the crack opening is de-
(Norling Mjonell
b
1997). Under this assumption and
2,3 creased
G1 (α c , α app. c for sthe same tcor comparing plain
28%
)= k α c + k α s (5)
0.06
by substituting b
Equation 1 into Equation 2 one
2,4 concrete swith vgfibre
c reinforced
vg s concrete for C=10mm.
obtains a Moreover, it is seen that in order to reach the same
Crack opening [mm]

0.05 2,2
a 2,3 crack kcvg and at
whereopening ksvgtheareconcrete
materialsurface,
parameters. From the
the thickness
∂w ∂h
a2,4
maximum amount of water per unit volume by
of corrosion products has to be increased thatapp.
can
e + ∇ • ( D ∇h) = ∂we ∂w
0.04

−
h α&c + e α&s + w&n (3) fill all pores (both capillary pores and gel pores),rein-
33% when comparing plain concrete with fibre one
∂h ∂t
0.03
∂α ∂α can calculate
forced K1 asThe
concrete. one obtains
thickness of the corrosion
c s products can be linked to time from numerical mod-
0.02
eling concerning evolution of ⎡a corrosion cell,
⎞ ⎤ e.g.
where ∂we/∂h is the slope of the sorption/desorption the modelw proposed by Michel ⎢
et
10 ⎜ g α − α ⎟h ⎥
⎛
al.
∞
c c
(2009) and
isotherm
0.01
(also called moisture capacity). The Equations 2-3.
− 0.188 α s + α s −G ⎢ −e
c 0.22
s 1
⎝1 ⎠
⎥

governing equation (Equation 3) must be completed ⎦ (6)

0 1
⎢ ⎥
0 K (The
α α effect of the increased ⎣
ductility of the mate-
by appropriate boundary
0.004 0.006 0.008
and initial conditions.
0.01 0.012 0.014 0.016 0.018 0.02 0.022 c s
, )=
⎜ g is
c − α c ⎟⎠h pronounced
⎛ ⎞
α more
∞
1
t [mm] rial, when adding fibres, when
The relation between the amount of evaporable
10

e
cor
⎝ 1
−
the cover thickness is increased to e.g. 40mm, cf.
1

water and relative humidity is called ‘‘adsorption

Figure 16. Crack opening at concrete surface for C=40mm and
various approaches for the cohesive relationship. Figure 12. When the crack isc openings the thickness
isotherm” if measured with increasing relativity The material parameters k to and
of the corrosion products has vg
k vg and
be app.
g1 can
50% higher
humidity and ‘‘desorption isotherm” in the opposite be calibrated by fitting experimental
for fibre reinforced concrete, approach a2,4, than for
data relevant to
4case. Neglecting their difference (Xi et al. 1994), in
DISCUSSION free (evaporable) water content in concrete at
plain concrete, approach b2,1, in order to obtain the
the following, ‘‘sorption isotherm” will be used with various ages (Di Luzio & Cusatis 2009b).
same crack opening at the concrete surface.
reference to the
Concerning bothcohesive
sorption relationships
and desorptionused conditions.
for the From Figure 13 it is seen that the mechanical per-
By the way, if the hysteresis of
simulations it is seen, that the variation mainly con- the moisture 2.2 Temperature
formance of a 10mm evolution
cover of FRC with cohesive
isotherm
cerns would be
the second branch takenofintothe account,
cohesive two different
relationship. parameters as material a2,4 is very similar to the me-
The first branch of the cohesive relationship must
relation, evaporable water vs relative humidity, con- Note that, at early age, since the chemical reactions
chanical performance of a 20mm cover of plain con-
be usedtheaccording
cerns concreteto the sign of
matrix, the variation
whereas of the
the second associated with cement hydration and SF reaction
crete with mechanical properties as material b2,1. Fi-
relativityconcerns
branch humidity.
the amountThe shape of properties
of fibres, the sorption of are exothermic, the temperature field is not uniform
nally, from Figure 14 it is seen that the same
isotherm
the fibres for
etc.HPC
Löfgrenis influenced by many
et al. (2008). Thusparameters,
the varia- for non-adiabatic systems even if the environmental
mechanical effect can be obtained by applying 20
especially
tions of thethose that relationship
cohesive influence extent givenand rate of 1-2
in Tables the temperature is constant. Heat conduction can be
mm FRC (a2,4) instead of 40 mm plain concrete
chemical reactions and, in turn,
concern variation of the second branch in order to determine pore described in concrete, at least for temperature not
(b2,1).
structure andthepore
investigate size distribution
properties changed (water-to-cement
when fibres are exceeding 100°C (Bažant & Kaplan 1996), by
The influence of constructing the cohesive rela-
ratio, cement
added. Comparing chemical composition,
the ductility SF content,
of the assumed cohe- Fourier’s law, which reads
tionship from two different approaches, cf. Section
sive relationships, Tables 1-2, with values additives,
curing time and method, temperature, mix from the 3.1, is illustrated in Figures 15-16. It is seen that the
etc.). In the
literature, literature
Kazemi et al various
(2007), formulations
it is seen, thatcan be
these q = − λ ∇T
influence on the crack opening at the concrete sur- (7)
found to describe the sorption
reflects the properties of concrete with small isotherm of normal face from the different cohesive relationships is in-
concrete of(Xifibres
amounts et al.(0.51994).
vol-%However, in the present
- 1.5 vol-%). where with
creased q isthethe heat influx,
increase is the absolute
coverTthickness.
paper the semi-empirical expression
From Figure 9 it is seen, that when a certain proposed by
value temperature, and λ is the
Though the same mechanical properties of heat conductivity; in this
the
of the corrosion thickness is reached, the crackit
Norling Mjornell (1997) is adopted because cover can be obtained for smaller covers with FRC

Proceedings of FraMCoS-7, May 23-28, 2010

than larger covers of plain concrete it is noted, that REFERENCES J = − D ( h, T )∇h
changing the cover thickness influences the transport
of hazardous substances to the reinforcement. Thus Andrade, C. et al. 1993.The Coverproportionality coefficient
cracking as a function of bar D(h,T)
corrosion: Part I- Experimental test. Materials and Struc-
tures, 26:453-464.moisture permeability and it is a nonlinea
the initiation of corrosion might be advanced when
the cover is decreased. However in order to evaluate Alonso, C. et al. 1997.of Factors
the relative humidity
controlling h and
cracking temperature
of concrete
this issue, a simulation of the formation of corrosion & Najjarcorrosion,
affected by reinforcement 1972). The moisture
Materials mass balanc
and Struc-
cells on embedded rebars has to be performed, e.g. tures, 31: 435-441.that the variation in time of the water mas
like proposed by Michel et al. (2009). Caré, S., Nguyen, Q. volume of concrete
T., L´Hostis, (water content
V. & Berthaud,
Mechanical properties of the rust layer induced by im-
B. 2008.w) be eq
divergence of the moisture flux J
pressed current method in reinforced mortar. Cement and
Concrete Research 38: 1079-1091.
5 CONCLUSION Chernin, L. et al. 2009.∂Analytical modelling of concrete cover
− wcorrosion
= ∇ • J of
cracking caused by ∂t reinforcement. Materials
A numerical model for the initiation and propagation and Structures published online
of cracks in a semi-infinite concrete body has been Hillerborg, A. et al 1976. Analysis of crack formation and
The water content w canmechanics
be expressed a
presented. The model is based on Hillerborgs ficti- crack growth in concrete by means of fracture
of the evaporable water
and finite elements. Cement and Concrete Research, w e (capillary
6: wa
tious crack model where MODE I cracking and 773-782. vapor, and adsorbed water) and the non-e
Mixed mode cracking is taken into account. The (chemically
Isgor, B. O. & Razaqpur, G. A. 2006. bound)
Modellingwater wn (Mil
steel corro-
cracking of the cover is modeled by the use of inter- sion in concrete Pantazopoulo
structures. Materials & Mills
and 1995).
Structures It
39: is reas
face elements with a cohesive relationship and the 291–302.
assume that the evaporable water is a fu
expansion of the corrosion products is modeled with Kazemi, M. T., et al. 2007, Cohesive Crack Model and Frac-
a fictitious thermal load. A uniform distribution of ture Energy of relative humidity, h, degreeNotched
Steel-Fiber-Reinforced-Concrete of hydration
degreeJournals
Cylindrical Specimens, of silicaof fume reaction,
Materials , i.e. we=w
in CivilαsEn-
the corrosion products is assumed. gineering 19(10):=884-890.age-dependent sorption/desorption
Based on the numerical simulation of the crack Liu, Y. & Weyers, R.E. (Norling Mjonellthe1997).
1998. Modeling Under this assum
Time-to-Corrosion
initiation and propagation in concrete covers due to by substituting Equation 1Concrete
Cracking in Chloride Contaminated Reinforced into Equati
reinforcement corrosion it is concluded: Structures, ACI Materials Journal 95(6): 675-681.
obtains
• For the cohesive relationships investigated, the Löfgren, I., et al. 2008. The WST method, a fracture mechanics
test method for FRC, Materials and Structures, 47: 197-
crack opening at the steel-concrete interface is 211. ∂w ∂h ∂w ∂w
independent of the ductility of matrix. Michel, A., et al. 2009 e + ∇Formation
− Modeling • ( D ∇h)of= Crackse α in Con-
&+ &+eα w
• After a certain thickness of the corrosion prod- ∂ h ∂t h ∂α c
∂α s

ucts is reached, the crack propagation is app.

crete Cover due to Reinforcement Corrosion, FraMCos-7
7th International Conference on Fracture Mechanics of
c -s

linear with tcor. Concrete and Concrete Structures; Conference proceed-

• The crack opening at the concrete surface is al- ings. Jeju, Korea,where
23-28 May /∂h is the slope of the sorption/
∂we2010.
most independent of the mechanical proper-
isotherm
Molina, F.J. et al. 1993. (also ascalled
Cover cracking moisture
a function of rebar capac
corrosion: Part 2 governing
– Numericalequation (Equation
model, Materials 3) must be
and Struc-
ties of the matrix when regarding a small tures, 26:532-548.by appropriate boundary and initial conditi
cover (10mm). However, as the cover thick-
ness is increased the influence of the material
The relation between the amount of e
Noghabai, K. 1999. FE_Modelling of cover splitting due to
corrosion by use of inner softening band. Materials and
properties gets more pronounced, and in- Structures, 32: 486-491. and relative humidity is called ‘‘
water
creased ductility of the material leads to de-
isotherm”
Rosello et. al 2005. Fracture if measured
of model with increasing
concrete: 2. Fracture en-
creased crack opening at the concrete surface.
humidity
ergy and characteristic andCement
length, ‘‘desorption isotherm”
and Concrete Re- in th

Furthermore the crack propagates to the con-

Val, D. V., Chernin,the L. & following,
crete surface for small covers (C=10mm) and
small values of tcor whereas the crack propa-
gation is stable inside the concrete cover for
larger cover thicknesses (C=40mm).
• The mechanical performance of a 20mm cover
of plain concrete, b2,1, is within the same
range as the mechanical performance of FRC,
a2,4. The same can be seen for 40mm plain
concrete, b2,1, and 20mm FRC, a2,4.
ACKNOWLEDGEMENTS
The first author is grateful to COWI A/S and The
Danish Agency for Science, Technology and Inno-
vation for the full support of the PhD project “Ap-
plication of Fibre Reinforced Concrete in Civil In-
frastructure” for Anders Ole Stubbe Solgaard.
case. Neglecting their difference (Xi et al.
search, 36:1345-1353.
Stewart, M.‘‘sorption isotherm” will be
G. 2009. Experimental
and numerical investigation of corrosion-induced cover
reference to both sorption
cracking in reinforced concrete structures. Journal and desorption c
of Struc-
By the
tural Engineering 135:376-385. way, if the hysteresis of the
isotherm would be taken into account, two
relation, evaporable water vs relative humi
be used according to the sign of the varia
relativity humidity. The shape of the
isotherm for HPC is influenced by many p
especially those that influence extent and
chemical reactions and, in turn, determ
structure and pore size distribution (water-
ratio, cement chemical composition, SF
curing time and method, temperature, mix
etc.). In the literature various formulatio
found to describe the sorption isotherm
concrete (Xi et al. 1994). However, in th
paper the semi-empirical expression pro
Norling Mjornell (1997) is adopted b

Proceedings of FraMCoS-7, May 23-28, 2010