Fracture Mechanics of Concrete and Concrete Structures -

Recent Advances in Fracture Mechanics of Concrete - B. H. Oh, et al.(eds)

Analysis of the dam-foundation joint through the cohesive frictional crack
2010 Korea Concrete Institute, Seoul, ISBN 978-89-5708-180-8

model
F. Barpi & S. Valente
Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, Torino.
Analysis of the dam-foundation joint through the cohesive frictional crack
model
F. Barpi & S. Valente
Dipartimento di Ingegneria Strutturale e Geotecnica, Politecnico di Torino, Torino.

ABSTRACT: The mechanical behaviour of joints plays a key role in concrete dam engineering since the joint
is the weakest point in the structure and therefore the evolutionary crack process occurring along this line
determines the global load bearing capacity. The reference volume involved in the above mentioned process is so
large that it cannot be tested in a laboratory: a numerical model is needed. The use of the asymptotic expansions
proposed by Karihaloo and Xiao 2008 at the tip of a crack with normal cohesion and Coulomb friction can
overcome the numerical difculties that appear in large scale problems when the Newton-Raphson procedure is
applied to a set of equilibrium equations based on ordinary shape functions (Standard Finite Element Method).
In this way it is possible to analyse problems with friction and crack propagation under the constant load induced
by hydromechanical coupling. For each position of the ctitious crack tip, the condition K1 = K2 = 0 allows
us to obtain the external load level and the tangential stress at the tip. If the joint strength is larger than the value
obtained, the solution is acceptable, because the tensile strength is assumed negligible and the condition K1 = 0
is sufcient to cause the crack growth. Otherwise the load level obtained can be considered as an overestimation
of the critical value and a special form of contact problem has to be solved along the ctitious process zone. For
the boundary condition analysed (ICOLD benchmark on gravity dam model), after an initial increasing phase,
the water lag remains almost constant and the maximum value of load carrying capacity is achieved when the
water lag reaches its constant value.

1 INTRODUCTION tip is lumped into a discrete line (two-dimensional)

The mechanical behaviour of joints plays a key role in
concrete dam engineering since the joint is the weak-
est point in the structure and therefore the evolution-
ary crack process occurring along this line determines
the global load bearing capacity. In the scientic liter-
ature two problems are discussed:
the problem of sliding along a pre-existing com-
pressed discontinuity (see, among others, Gens,
Carol, and Alonso 1990),
the problem of crack initiation and propagation
along an undamaged interface (see Carol, Prat,
and Lopez 1997, Cervenka, Kishen, and Saouma
1998, Barpi and Valente 2008).
The latter problem is discussed below in the frame-
work of the cohesive crack models, introduced by
Barenblatt and Dugdale for elastoplastic materials,
and by Hillerborg et al. for quasi-brittle materials. In
this model, the nonlinear fracture process zone (due
to degradation mechanisms such as plastic micro-
voiding or micro-cracking) in front of the actual crack
or plane (three-dimensional) and is represented by
a traction-separation law across this line or plane.
When the tangential components of the tractions are
present the solution can lose uniqueness. Therefore
numerical difculties occur if the Newton-Raphson
procedure is applied to a set of equilibrium equations
based on ordinary shape functions (Standard Finite
Element Method). In order to overcome these dif-
culties Strouboulis, Copps, and Babuska 2001 sug-
gest constructing an approximation which employs
knowledge about the character of the solution (Gen-
eralized Finite Element Method). In this direction we
take advantage from the work of Karihaloo and Xiao
2008 on the asymptotic elds at the tip of a cohesive
crack. In this model frictional forces operate when the
crack faces are open. Therefore these forces are differ-
ent from those operating in a contact problem. In this
context Karihaloo and Xiao 2008 obtained asymptotic
expansions at a cohesive crack tip analogous to the
Williams (1957) expansions at a traction-free crack
tip for any traction-separation law that can be ex-
pressed in a special polynomial form.
 J = D ( h , T ) h
2 POLYNOMIAL COHESIVE LAW FOR QUASI-
BRITTLE MATERIALS The proportionality coefficient D(h,T)
moisture permeability and it is a nonlinea
1
of the relative humidity h and temperature
& Najjar 1972). The moisture mass balanc
0.8
that the variation in time of the water mas

Non dimensional stress (-)

volume of concrete (water content w) be eq
0.6
divergence of the moisture flux J

= J
0.4 w

t
0.2

The water content w can be expressed a

of the evaporable water we (capillary wa
0

vapor, andopening
adsorbed water) and the non-e
0 0.2 0.4 0.6 0.8 1
Figure 1. Stresses near the crack tip. Non dimensional (-)

(chemically bound) water wn (Mil

Pantazopoulo
Figure 2. Non dimensional & Mills
opening (w/w 1995).
c ) vs. non It is reas
dimen-
In order to obtain the separable asymptotic eld at sional stress (/c ).assume that the evaporable water is a fu
a cohesive crack tip (in terms of r and functions, see relative humidity, h, degree of hydration
Fig. 1) in quasi-brittle materials, Karihaloo and Xiao degree of silica fume reaction, s, i.e. we=w
2007 reformulate the softening law into the following where a prime denotes = age-dependent
differentiation sorption/desorption
with respect
polynomial form: (Norling
to z and an overbar complex Mjonell 1997). In
conjugate. Under
Eq. this
4, assum
  2i3   4 = E/[2(1 + )] by substituting
is the shear modulus;Equation 1 into Equati
the Kolosov
y
5
w
5
w obtains
=1+ i 1+ i constant is = 3 4 for plane strain and = (3
ft i=1
wc i=1
wc )/(1 + ) for plane stress; E and are Youngs mod-
h + ( D h) = we & + we & + w
w respectively.
e mode
(1) ulus and Poissons ratio,
where y and ft are the stress normal to the cohe- For a general mixedh t I+IIhproblem,
thec two
s

sive crack face and the uniaxial tensile strength, re- c

analytic functions (z) and (z) can be chosen as se- s
spectively; w and wc are the opening displacement ries of complex eigenvalue Goursat functions (Sih and
of the cohesive crack faces and the critical displace- Liebowitz 1968) where we/h is the slope of the sorption/
ment at the real crack tip; i are tting parameters. isotherm (also called moisture capac
Equation 1 can represent a wide variety of softening governing equation (Equation 3) must be



by appropriate boundary and initial conditi
laws. For example, Karihaloo and Xiao 2007 showed (z) = An z = An r ei n n n

that the experimental results of Cornelissen, Hordijk, The relation between the amount of e
and Reinhardt 1986 for normal concrete can be tted
n=0
watern=0 and relative humidity is called
very well by Eq. 1 with:1 = 0.872, 2 = 16.729,
isotherm
if measured with increasing
(z) = z +1 = andAdesorption
Bnhumidity nr
+1 i( +1)
e isotherm
n
(5) in th n n
3 = 67.818, 4 = 110.462, 5 = 83.158 (see Fig.
2). The above mentioned shape coefcients are used
n=0 case. Neglecting their difference (Xi et al.
n=0

in the present work. the following, sorption isotherm will be

where the complex reference to both
coefcients An = a1nand
are sorption + desorption
ia2n c
and Bn = b1n + ibBy The eigenvalues
2n the way, if the n hysteresis of the
. and coef-
3 ASYMPTOTIC FIELDS AT THE TIP OF A cients a1n , a2n , b1nisotherm
and b2n are
CRACK WITH NORMAL COHESION AND would
real.be taken into account, two
COULOMB FRICTION relation, evaporable
Substituting complex functions water
5 into 2, 3vsand
relative
4, humi
The mathematical formulation follows closely that
the complete series be used according to the sign of the varia
expansions of the displacements
used by Karihaloo and Xiao 2008, so only a brief de-
and stresses near the relativity
tip of thehumidity.
crack can The shape of the
be written:
scription will be given here. Muskhelishvili (1953)
isotherm for HPC is influenced by many p
showed that, for plane problems, stresses and dis-
 especially those that influence extent and
placements in the Cartesian coordinate system (see 2u = r chemical

k(a1n cos reactions
n
nand, in turn, determ
e.g. Fig. 1) can be expressed in terms of two ana- n=0 structure and pore size distribution (water-
lytic functions (z) and (z) of the complex variable
ratio, cement chemical composition, SF
z = rei a2n sin n )curing time1nand
+ n [a cos(method,
n 2)+
temperature, mix
etc.). In the literature various formulatio
x + y = 2[
(z) +
(z)] (2) a2n sin(n 2) found ton describe
+ ( + 1)(b1nthe cossorption
n + isotherm
concrete (Xi et al. 1994). However, in th
y x + 2ixy = 2[z

(z) +

(z)] (3) paper the semi-empirical expression  pro
2(u + iv) = k(z) z
(z)
(z) (4)
Norling Mjornellb2n (1997)sin n ) is (6)
adopted b

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

J = D ( h , T ) h 

(1) explicitly accounts for the evolution of hydration
2v = rn k(a1n sin n reaction
The and SF
imposition content. This
of continuity sorptionon isotherm
conditions normal
The proportionality
n=0 coefficient D(h,T) is called reads
stress component of Eq. 9 (y |= = y |= ) along
moisture permeability and it is a nonlinear function the cohesive zone gives:
of the relative humidity
a2n cos n ) + nh[aand temperature
1n sin( n 1)+T (Baant (a2n + b2n ) sin(

n1 ) = 0

(13)
& Najjar 1972). The moisture mass balance requires
we (h, c , s ) = G1 ( c , s )1 1
+
that the
a2n variation
cos(n in 2)time
+ (of the water mass per unit
n + 1)(b1n sin n +
The imposition of continuity 10 (g
on)h tangen-
conditions
1 c c
volume of concrete (water content w) be equal to the e (4))
tial stress component of Eq. 10 (xy |= = xy |=

divergence of the moisture flux J 
b2n cos n )
along the cohesive zone gives:
g )h
10(
1 c c 1
(7)
K1(
[(n 1)a1n + ( , )b)e ] sin(
cn+1 s 1n n 1) = 0 (14)
= J
w
(2)

Equations 13 and 14 are satised for sin(n 1) = 0



t

or for b2n = a2n . In other words the asymptotic so-
x =
r n 1 2n [a1n cos(n 1)
The water content w can be expressed as the sum where can
lutions the befirst term (gel
collected isotherm)
in two represents
classes. The the
rst class
of the n=0
evaporable water we (capillary water, water physically bound (adsorbed) water
is characterized by integer eigenvalues: and the second
vapor, and nadsorbed term (capillary isotherm) represents the capillary
a2n sin( 1)] water)
(n and
1)[athe non-evaporable
cos( 3) water.
n = nThis
+ 1, expression
n = 0, 1, 2is. .valid
., w only
= 0,for low
= 0content
(15)
(chemically bound) nwater wn 1n (Millsn 1966, of SF. Theclass
coefficient G1 represents the remaining
amount of
Pantazopoulo & Mills 1995). It is reasonable to
a2n sin(n 3)] (n + 1)n [b1n cos(n 1)
the second
water per unit
is characterized
volume held in the
by the
gel pores at 100%
assume that the evaporable water is a function of cases (non integer eigenvalues):
relative humidity, and it can be expressed (Norling
relative humidity, h, degree of hydration, c, and
degree of silica fume reaction, s, i.e.nw
b2nsin( e=1)]
we(h,c(8)
,s) bMjornell 1997)b1nas = n 1 a1n , w
= 0,
= 0
2n = a2n ,
= age-dependent sorption/desorption isotherm n + 1
(Norling Mjonell 1997). Under this assumption and G c s
( ) = k c + k s
(16)
(5)
by substituting

Equation 1 into Equation 2 one 1 c s vgofcthe Coulombian
,
The imposition vg s friction condition
1 (xy |= = f y |= ) along the cohesive zone, for
obtains
y = r n
2n [a1n cos(n 1)
n=0 where
the c
and
rst kclass
vg
s
are material
of ksolutions gives: parameters. From the
vg
w w w maximum amount of water per unit volume that can
h
e sin( e n&+1)[ae1n&cos(

a2n
h t
( D1)]
+ n h)+= n (
h
& 3)(3)
s + wnn
fill
nall
= pores
n + 1 (both capillary pores and gel pores), one
c
c
s can calculate K1 as one obtains
a2n sin(n 3)] + (n + 1)n [b1n cos(n 1) na2n + (n + 2)b2n = f (n + 2)(a1n + b1n )
where we/h is the slope of the sorption/desorption 

g h

c c

10

isotherm (also called moisture

b2n sin(capacity).
1)] The
(9)
w s + s G n
c s
0.188 e
=

0,
0.22
1, 2 . .

.
(17) 1
1

governing equation (Equation 3) mustn be completed (6)

0 1

K (for
c sthe

by appropriate boundary and initial conditions. and1
, ) = second class of solutions gives:
g h

The
relation between the amount of evaporable c c

10

( a a e ) cos( 1)
= 0
1
(18)
f 1n 2n n 1

water
xy =and rrelative
n1
nhumidity
(n 1)[ais1n called
sin(n adsorption
3)
isotherm n=0 if measured with increasing relativity Theboth
Since material parameters
factors
vg and k vg and g1 can
in Eq. 18kc may s
vanish indepen-
humidity and desorption isotherm in the opposite be calibrated by fitting experimental data relevant
dently of each other, it appears that, for the crack withto
case.
a2nNeglecting
cos(n3 )] their difference
+ ( (Xi et al. 1994), in
n + 1)n [b1n sin(n 1) free (evaporable)
normal cohesion and water contentfriction,
Coulombian in concrete
the eigen-at
the following, sorption isotherm will be used with various ages (Di Luzio & Cusatis 2009b). Addi-
values and asymptotic elds are not unique.
reference to both sorption and desorption conditions.  tional assumptions have to be made to ensure unique-
By the way, if the hysteresis b 2n cos( 1)]
ofn the moisture
(10) ness. Assuming that f a1n a2n
= 0, Eq. 18 gives:
isotherm would be taken into account, two different 2.2 Temperature evolution2n + 3
relation, evaporable
 water vs relative humidity, must Note n 1)
cos(that, = 0, age,
at early n since
= , n = 0, 1, 2 . . .
2the chemical reactions
bew used
= v
 according
v
 to the sign of the variation of the
= associated with cement hydration and SF reaction (19)
relativity =humidity.
= The shape of the sorption are exothermic, the temperature field is not uniform
This assumption does not lead to any loss of general-
isotherm for HPC is influenced by many parameters, forNow
ity. non-adiabatic systems
it is possible even if the
to complete theexpressions
environmentalof
especially

r those that influence extent and rate of the
n temperature
the asymptoticiselds.
constant. Heat conduction can be
chemical [(kreactions
+ n )a1n and,
+ (inn +turn,
1)b1ndetermine pore
] sin n (11) described in concrete,
In the case of integerat eigenvalues,
least for temperature not
substituting
structure
n=0 and pore size distribution (water-to-cement exceeding 100C
Eq. 17 in 9 gives: (Baant & Kaplan 1996), by
ratio, cement chemical composition, SF content, Fouriers law, which reads
curing time and method, temperature, mix additives,
etc.).
= uIn theliterature
u = various formulations can be q=| T = xy |=
y = = (7)
found to=describe =the sorption isotherm of normal f
concrete

r
(Xi et al. 1994). However, in the present where
q is the heat flux, T is the absolute
paper the [(
n
semi-empirical expression proposed by
n k)a2n + (n + 1)b2n ] sin n (12) temperature,(nand is+the
+ 2)(n 1)rheat
n
(a1nconductivity;
b1n ) cos(n) in(20)
this
Norling
n=0
Mjornell (1997) is adopted because it n=1

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

 J = D ( h , T ) h
In the case of non-integer eigenvalues, substituting to be imposed. This special form of contact problem
Eqs. 16 and 19 in 11 and 12 gives: is beyond the scope ofThethe proportionality
present work. coefficient D(h,T)
moistureliterature
In the well established permeability and it is be-
on mechanical a nonlinea
   of the relative humidity h and
haviour of concrete joints (see Cervenka, Kishen, and temperature

r 2n+3
2 2n + 3 2n + 5 Saouma

& Najjar depends
1998), softening 1972). Theonlymoisture
on wefmass
f = balanc
w= + a1n + b1n that the variation in time of the water mas
2 2 w + . In the asymptotic expansion used, soften-
2 2

ing depends only on w. Therefore, during the iterative w) be eq

volume of concrete (water content
n=0

2n + 3
divergence
procedure, wc changes of the moisture flux J
as follows:
sin (21)
2 w
wci+1 = = 2 J ( i )2
t wef f,c (24)

r 2n+3    The water content w can be expressed a

2n + 3 2n + 5
2
= a2n + b2n
n=0
2 2
sin
2n + 3
(22)
2 assume that the evaporable water is a fu
In Eq. 19 n = 1 corresponds to the singular terms,
which are excluded a priori (K1 = K2 = 0).
4 THE ITERATIVE SOLUTION PROCEDURE
For each position of the ctitious crack tip (shortening
FCT) the following iterative procedure is applied:
 i+1   i 
w
=f
y c
xy

y
i+1   i+1 
w
xy
=g
i = 0, 1, 2 . . . (23)
Since the material outside the fracture process zone
(shortening FPZ) is linear, it is possible to compute
the external load multiplier () and the tangential
stress at the FCT (xy,F CT ) by imposing that the stress
eld is not singular (stress intensity factors K1 =
K2 = 0). All these linear constraints are included in
the operator f .
Since w,,y ,xy are compatible with the asymp-
totic solution, operator g includes the constraints de-
scribed in Karihaloo and Xiao 2008 and not repeated
here.
At the rst iteration (i = 0) w = = 0 is assumed
along the FPZ. According to this approach and
xy,F CT are not dened a priori but are obtained from
the analysis related to a pre-dened position of the
FCT. If xy,F CT is less than or equal to the local crit-
ical value, the solution obtained can be accepted. On
the contary, if xy,F CT exceeds the local critical value,
the associated load level can be seen as an overestima-
tion of the real critical value which remains unknown.
Of course it is possible to reduce the load level but
in that case K1 becomes negative, a contact problem
arises along the FPZ and the dilatancy condition has
5 NUMERICAL EXAMPLE
of the evaporable water we (capillary wa
vapor, and adsorbed water) and the non-e
(chemically bound) water wn (Mil
Pantazopoulo & Mills 1995). It is reas
relative humidity, h, degree of hydration
degree of silica fume reaction, s, i.e. we=w
= age-dependent sorption/desorption
(Norling Mjonell 1997). Under this assum
by substituting Equation 1 into Equati
obtains
w h
e + ( D h) = we w
&c + e &s + w
h t h
s
Figure 3. Failure criterion.
where we/h is the slope of the sorption/
isotherm (also called moisture capac
governing equation (Equation 3) must be
As an example of byapplication,
appropriatetheboundary and initial
benchmark prob- conditi
lem proposed in 1999 Theby relation between Commis-
the International the amount of e
water and relative humidity
sion On Large Dams ICOLD 1999 was analysed (dam is called
isotherm if
height 80m, base 60m, see Fig. 4). measured with increasing
humidity and desorption isotherm
For simplicity the same value of Youngs modulus
in th
case. Neglecting their difference (Xi et al.
(E = 32.5 GPa) and Poissons ratio ( = 0.125) was
the following, sorption isotherm will be
assumed. Figure 3 shows the Mohr envelope of peak
reference to both sorption and desorption c
and residual strength for the joint (cohesion=0.7MPa,
By the way, if the hysteresis of the
= 30o ). The stress x is positive (tension) along
isotherm would be taken into account, two
the lower edge of the crack. Figure 3 shows its con-
relation, evaporable water vs relative humi
tribution to the achievement of the critical condition.
be used according to the sign of the varia
As the crack grows, the value of x at the FCT (also
relativity humidity. The shape of the
called T-stress) reduces. For conservative reasons, the
isotherm for HPC is influenced by many p
tensile strength of the joint and the related fracture en-
especially those that influence extent and
ergy are assumed as negligible. In case of linear soft-
chemical reactions and, in turn, determ
ening the ICOLD benchmark suggests the assumption
structure and pore size distribution (water-
of a critical value of the crack sliding displacement
ratio, cement chemical composition, SF
equal to c = 1 mm. Since the shape of the soften-
curing time and method, temperature, mix
ing law assumed in the present paper is based on the
etc.). In the literature various formulatio
results of Cornelissen, Hordijk, and Reinhardt 1986,
found to describe the sorption isotherm
the previous value was increased to c =2.56 mm. This
concrete (Xi et al. 1994). However, in th
choice is motivated by keeping constant the fracture
II paper the semi-empirical expression pro
energy GF in the case w = 0. Since the crack is open,
Norling Mjornell (1997) is adopted b
beyond this value no stress transfer occurs.

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

 J = D ( h , T ) h (1) explicitly accounts for the evolution of hydration
reaction and SF content. This sorption isotherm
The proportionality coefficient D(h,T) is called

reads
moisture permeability and it is a nonlinear function
of the relative humidity h and temperature T (Baant

7

& Najjar 1972). The moisture mass balance requires we (h6 c s ) = G1 ( c , s )1 1
+
that the variation in time of the water mass per unit
, ,

c )h

(g
volume of concrete (water content w) be equal to the 5 e

c 10
1
(4)
divergence of the moisture flux J )h
 

(g
   

Water lag (m)

4
K ( c s ) e
, c 10
c 1
1

= J
   

(2)
1
w 3

t

where the first term (gel isotherm) represents the

2

The water content w can be expressed as the sum physically bound (adsorbed) water and the second
of the evaporable water we (capillary water, water
1

vapor, and adsorbed water) and the non-evaporable term (capillary isotherm) represents the capillary
(chemically bound) water wn (Mills 1966, water. This expression is valid only for low content
0
0 5 10 15 20

of SF. The coefficient G1 represents the amount of

Distance of FCT from the upstream edge (m)
Pantazopoulo & Mills 1995). It is reasonable to water per unit volume held in the gel pores at 100%
assume that the evaporable water is a function of relative humidity, and lag
Figure 5. Water it can be expressed
vs. FCT position. (Norling
relative humidity, h, degree of hydration, c, and Mjornell 1997) as
degree of silica fume reaction, s, i.e. we=we(h,c,s)

= age-dependent sorption/desorption isotherm

(Norling
Figure Mjonell
4. Gravity 1997).
dam Under
proposed this assumption
as benchmark and
by ICOLD G c
foregoing discussion,
( , ) = k c + k s
wes can conclude that the asso- (5)
by substituting Equation 1 into Equation 2 one
(1999). ciated
1 cload s level
vg hcovt shown
vg s in Fig. 7 is just an over-
estimation of the real level. This model behaviour is
obtains
5.1 Water lag where
due to the and kvalue
kcvg low s
vg are material parameters. From the
of cohesion suggested by the
Thewwell w w maximum amount of water
benchmark. For higher values perofunit volumethethat
cohesion can
solu-
e h established e & on
literature e & +driven
water
&
frac-
c +Lenoach,
(see + ( D hDetournay,
) = w (3) fill all
tion shownporesin(both
Fig. 8capillary
and 7 ispores and gelacceptable.
completely pores), one
ture h t
Desroches, h s Papanasta-
n
can calculate
Figure 7 givesKthe maximum value of hovt which is
1 as one obtains
siou, Pearson, Thiercelin, cand Cheng s 1994) assume also the maximum load carrying capacity of the dam.
that the water penetrates into the crack but does not Figure 8 shows the evolution of the horizontal crest
reach e/h is
wherethewFCT. Thethefraction
slope ofoftheFPZ sorption/desorption
not reached by displacement g
h

c c and
w as a function
s + s sof the FCT position 10

isotherm
the water is(also
calledcalled moisture
water lag. Accordingcapacity).
to the ex-The Fig. 9 the deformedcmesh along
G e
the joint.
0.188 0.22 1
1

governing results
equationof (Equation
Reich, Bru3)hwiler,
must be completed (6)
0 1

perimental Slowik, and K ( c s ) =
by appropriate
Saouma 1994, itboundary
is assumed andthat
initial
the conditions.
water penetrates 1
,

g h

intoThethe relation
FPZ up between the amount knee of evaporable e c c
10
1
to the conventional point of 1

water
the and relative
softening law (w > humidity
wef f,c is2/9called
= 2.56 adsorption
2/9 =
isotherm
0.569 mm.) At if the
measured with the
points where increasing relativity
water penetrates, The material parameters kcvg and ksvg and g1 can
humidity and desorption isotherm
the pressure is the same as in the reservoir at the in the opposite
same be calibrated by fitting experimental data relevant to
case. Neglecting
depth. The concrete their
anddifference
the rock are(Xi assumed
et al. 1994), to bein free (evaporable) water content in concrete at
1.5

the following,
impervious. Thesorption
asymptotic isotherm
expansion willusedbe used with
is based various ages (Di Luzio & Cusatis 2009b).
reference to both sorption
xy |= = anddesorption
| conditions.
Tangential stress ratio at FCT (-)

on the assumption f y = therefore it

By the way, if the hysteresis of
can be applied only in the region not reached by the the moisture 2.2 Temperature evolution
isotherm would be taken into account, twowater
different
1.4
water. Figure 5 shows the evolution of the lag
asrelation, evaporable
a function of the FCT water vs relative humidity, must
position. Note that, at early age, since the chemical reactions
be used according to the sign of the variation of the associated with cement hydration and SF reaction
relativity
5.2 Loading humidity.
conditions The shape of the sorption are exothermic, the temperature field is not uniform
isotherm
dam isfor HPC isunder
influenced by many parameters,
1.3

The analysed self-weight, reservoir ll- for non-adiabatic systems even if the environmental
especially those that influence extent and
ing and imminent failure ood loading conditions. rate of the
In temperature is constant. Heat conduction can be
chemical
the numerical reactions
analysisand, in turn,
the role determine
of external pore
load mul- described in concrete, at least for temperature not
structure
tiplier wasand porebysizethedistribution
played water level(water-to-cement
above the dam exceeding 100C (Baant & Kaplan 1996), by
1.2

ratio,alsocement chemical composition,

water heigth SF content, Fouriers law, which reads
0 5 10 15 20
Distance of FCT from upstream edge (m)
crest called overtopping (shortening
curing time and method, temperature, mix additives,
hovt = hif f hc , see Fig. 4). Under the conservative
etc.). In thepreviously
assumptions literature described
various formulations
related to the can
mate-be q = 6.Tangential
Figure T (7)
stress ratio xy /cohesion at FCT vs.
found to describe the sorption isotherm of
rial properties, the crack starts before the water level normal FCT position.
concretethe(Xi
reaches dametcrest
al. 1994).
(hovt <However,
0). in the present where q is the heat flux, T is the absolute
paper the semi-empirical expression proposed by
Figure 6 shows the evolution of ( /cohesion) at the temperature, and is the heat conductivity; in this
Norling Mjornell (1997) is adopted
FCT as a function of the FCT position. Based on the
because it

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

 J = D ( h , T ) h
6 CONCLUSIONS
The proportionality
The reference volume coefficient
involved in the fracture D(h,T)
process of a dam joint is so large that ititcannot
moisture permeability and is a nonlinea
6
of the relative humidity h and
be tested in a laboratory: a numerical model temperature
is
needed. & Najjar 1972). The moisture mass balanc
that the variation in time of the water mas
volume of concrete (waterproposed
content w) be eq
4

Overtopping height (m)

The use of the asymptotic expansions

divergence of the moisture
by Karihaloo and Xiao 2008 at the tip fluxofJ a
2
crack with normal cohesion and Coulomb fric-
w =the
tion can overcome numerical
J
difculties that
0 t problems when the Newton-
appear in large scale
Raphson procedure is applied to a set of equi-
librium equationsThe water
based on content can befunc-
ordinarywshape expressed a
-2
of Finite
tions (Standard the evaporable water we (capillary wa
Element Method).
vapor, and adsorbed water) and the non-e
In this way it(chemically bound)
analysewater wn (Mil
0 5 10 15 20
Distance of FCT from upstream edge (m) is possible to problems
with friction Pantazopoulo & Mills 1995).
and crack propagation It is reas
under the
Figure 7. Overtopping height hovt vs. FCT position. constant loadassume
inducedthat the evaporable water
by hydromechanical cou- is a fu
pling. relative humidity, h, degree of hydration
degree of silica fume reaction, s, i.e. we=w
In the analysis= of the
age-dependent
dam-foundation sorption/desorption
joint pene-
trated by the water, for each position ofUnder
(Norling Mjonell 1997). this assum
the FCT,
the conditionby substituting
K1=K2=0 allowsEquation 1 into
us to obtain the Equati
external loadobtains
level and the tangential stress at
the FCT. If the joint strength is larger than the
w
value obtained,wthehsolution is acceptable, w
Horizontal crest displacement (cm)

e & +be-
1
e + is( Dassumed
strength h) = negligible e & + w
cause the tensile
h t h c
s
and the condition K1 = 0 is sufcient cto cause s
the crack growth. Otherwise the load level ob-
tained can bewhere we/h
considered as is
anthe slope of the ofsorption/
overestimation
0.5
isotherm
the critical value and a (also
specialcalled
form ofmoisture
contact capac
problem havegoverning equation
to be solved along the(Equation
FPZ. 3) must be
by appropriate boundary and initial conditi
For the boundary Thecondition
relation analysed,
between the afteramount
an of e
initial increasingwaterphase,
and therelative humidity
water lag remainsisal- called
most constant.isotherm if measured with increasing
0
0 5 10 15 20
Distance of FCT from upstream edge (m)
humidity and desorption isotherm in th
For the boundary case. condition
Neglectinganalysed,
their difference
the max-(Xi et al.
imun value ofthe following,
carrying sorption
capacity isisotherm
achieved will be
Figure 8. Horizontal crest displacement vs. FCT position.
load
when the water reference to both
lag reaches sorption value.
its constant and desorption c
By the way, if the hysteresis of the
7 ACKNOWLEDGMENTS isotherm would be taken into account, two
The nancial support relation, evaporable
provided by thewater vs relative
Italian Min- humi
be used according to
istry of Education, University and Scientic Researchthe sign of the varia
(MIUR) to the research project on Structural mon- of the
relativity humidity. The shape
itoring, diagnosticisotherminverse for HPC isand
analyses influenced
safety as-by many p
sessments of existing concrete dams (grant number and
especially those that influence extent
20077ESJAP 003)chemical reactions
is gratefully and, in turn, determ
acknowledged.
structure and pore size distribution (water-
REFERENCES ratio, cement chemical composition, SF
curing (2008).
Barpi, F. and S. Valente time and method,
Modeling temperature,
water penetra- mix
etc.). Injoint.
tion at dam-foundation the Engineering
literature various formulatio
Fracture Me-
Figure 9. Deformed mesh.
chanics 75/3-4, found to describe
629642. the sorption
Elsevier Science isotherm
Ltd. (Great
Britain). concrete (Xi et al. 1994). However, in th
paper
Carol, I., P. Prat, and the(1997).
C. Lopez semi-empirical
A normal/shear expression
crack- pro
Norling
ing model: Application toMjornell
discrete (1997)
crack is
analysis.
nal of Engineering Mechanics (ASCE) 123(8), 765773.
adopted
Jour- b

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

J = D ( h , T ) h (1) explicitly accounts for the evolution of hydration
Cornelissen, H., D. Hordijk, and H. Reinhardt (1986). Exper- reaction and SF content. This sorption isotherm
Theimental
proportionality
determination coefficient D(h,T)
of crack softening is called
characteristics reads
moisture permeability and it is a nonlinear function
of normal and lightweight concrete. Heron 31, 4556.
ofDesroches,
theJ.relative
J., E.humidity
Detournay, h and temperature
B. Lenoach, T (Baant
P. Papanastasiou,
& Najjarcrack1972).
tip region The moisturefracture.
in hydraulic mass balance
Pearson, M. Thiercelin, and A. Cheng (1994). The
Proceedings requires
of the we (h c s ) = G1 ( c , s )1 1
+
that the variation
Society ofinLondon
time Aof447,
the3948.
water mass per unit
, ,

Royal (g
c 10 c )h
volume
Gens, A.,of concrete
I. Carol, and (water content(1990).
E. Alonso w) beAequal to the
constitutive e

1
(4)
divergence
model forof rock
the moisture flux J and numerical imple-
joints, formulation (g
)h
mentation. Computers and Geotechnics 9, 320.
K ( c s ) e
, c c
10
1
1

w =concrete
ICOLD (1999). Theme A2: Imminent failure ood for a
(2)
1
J
t Workshopgravity dam. In Fifth International Benchmark
on Numerical Analysis of Dams, Denver (CO).

Thefrictionless
water content w can be expressed
Karihaloo, B. and Q. Xiao (2007). Accurate simulation of
as inthequasi-
sum where the first term (gel isotherm) represents the
of thebrittle
evaporable
and frictional
water w
cohesive crack growth
(capillary water, water physically bound (adsorbed) water and the second
vapor,barova,
materials using xfem.
and adsorbed water)
e In A. Carpinteri, P. Gam-
and the(Eds.),
non-evaporable
Sixth Inter-
term (capillary isotherm) represents the capillary
G. Ferro, and
Conference on
G. Plizzari
(chemically bound) water wn (Mills of 1966,
national Fracture Mechanics Con- water. This expression is valid only for low content
Pantazopoulo & Mills 1995). It is reasonable to
crete and Concrete Structures (FRAMCOS6), pp. 99 of SF. The coefficient G1 represents the amount of
assume that the evaporable water is a function of
110. Taylor and Francis (London). water per unit volume held in the gel pores at 100%
relative humidity, , degree of hydration, , and
cFrac- relative humidity, and it can be expressed (Norling
a cohesivehcrack.
Karihaloo, B. and Q. Xiao (2008). Asymptotic elds at the
tip of
degreetureof150,
silica
International Journal of
fume reaction, s, i.e. we=we(h,c,s) Mjornell 1997) as
5574.
= Reich,
age-dependent
W., E. Bruhwiler,sorption/desorption isotherm
(Norling Mjonell and
Experimental 1997).
V. Slowik, and V. Saouma
Under thisaspects
computational assumption
(1994).
of a wa- and c c+ ks s
G ( c s ) = k vg (5)
by substituting Equation 1 into Equation 2 one
ter/fracture interaction. The Netherlands, pp. 123131. 1
,
c vg s
obtains Balkema.
Sih, G. and H. Liebowitz (1968). Mathematical theories of where kcvg and ksvg are material parameters. From the
brittle fracture. In H. Liebowitz (Ed.), Fracture (vol. II),
w pp. maximum amount of water per unit volume that can
e h67190. Academicw e (NewwYork).
e

h t
+ ( D h ) =
T., K.hCopps, and
&+
Press
& + w& (3) fill all pores (both capillary pores and gel pores), one
Strouboulis, I. Babuska
c
c
(2001).
s
s

alized nite element method. Computer Methods in Ap-

n
The gener- can calculate K1 as one obtains
plied Mechanics and Engineering 190, 40814193.
where
we/h
Cervenka, J., J. is the slope
Kishen, and V.of the sorption/desorption
Saouma (1998). Mixed mode

10

g c c h
isothermfracture(also called moisture
of cementious capacity).partThe
bimaterial interfaces; ii: w 0.188 s + 0.22 s G
c s 1 e 1

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

0 1

Numerical simulations. Engineering Fracture Mechan-
K ( c s ) =
by appropriate boundary and initial conditions.
ics 60(1), 95107. 1
,

g c c h

The relation between the amount of evaporable

10
e 1
1

water and relative humidity is called adsorption

isotherm if measured with increasing relativity
humidity and desorption isotherm in the opposite
case. Neglecting their difference (Xi et al. 1994), in
the following, sorption isotherm will be used with
reference to both sorption and desorption conditions.
By the way, if the hysteresis of the moisture
isotherm would be taken into account, two different
relation, evaporable water vs relative humidity, must
be used according to the sign of the variation of the
relativity humidity. The shape of the sorption
isotherm for HPC is influenced by many parameters,
especially those that influence extent and rate of the
chemical reactions and, in turn, determine pore
structure and pore size distribution (water-to-cement
ratio, cement chemical composition, SF content,
curing time and method, temperature, mix additives,
etc.). In the literature various formulations can be
found to describe the sorption isotherm of normal
concrete (Xi et al. 1994). However, in the present
paper the semi-empirical expression proposed by
Norling Mjornell (1997) is adopted because it
The material parameters kcvg and ksvg and g1 can
be calibrated by fitting experimental data relevant to
free (evaporable) water content in concrete at
various ages (Di Luzio & Cusatis 2009b).
2.2 Temperature evolution
Note that, at early age, since the chemical reactions
associated with cement hydration and SF reaction
are exothermic, the temperature field is not uniform
for non-adiabatic systems even if the environmental
temperature is constant. Heat conduction can be
described in concrete, at least for temperature not
exceeding 100C (Baant & Kaplan 1996), by
Fouriers law, which reads
q = T (7)
where q is the heat flux, T is the absolute
temperature, and is the heat conductivity; in this

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