Sunteți pe pagina 1din 13

Concrete fracture models

1. Early history

Initiated in 1961 by Kaplan, the study of fracture mechanics has progressed by
the turn of the century quite far. Kesler et al.(1972) showed that the classic linear
fracture mechanics (LEFM) of sharp cracks was inadequate for normal concrete
structures. This conclusion was supported by the results of Walsh (1972), who
tested geometrically similar notched beams of different sizes and plotted the
results in a double logarithmic diagram of normal strength versus size. Without
attempting a mathematical description, he made the point that this diagram
deviates form a straight line of slope -1/2 predicted by LEFM.
A major advance in concrete fracture was made by Hillerborg et al. (1976).
Inspired by the softening and plastic models of fracture process zone (FPZ)
initiated in the works of Barenblatt (1962) and Dugdale (1990) and developed
earlier for materials other than concrete by Rice (1968), Smith (1974), Knauss
(1973), Wnuk (1974), Palmer and Rice (1973), and Kfouri and Rice (1977).
Hillerborg et al. (1976) improved and adapted to cohesive crack model. Their
finite element analysis showed that the cohesive crack model (also called the
fictitious crack model) predicts, for the flexural failure of un-notched plain
concrete beams, a deterministic size effect, different from the Weibull statistical
size effect. This conclusion was strengthen and the model further refined by
Petersson (1981).
An analytical study of the size effect due to localization of distributed cracking
was begun by Bazant (1976). Later, the crack band model providing an almost
equivalent alternative to the cohesive crack model, was developed. This model
was shown to be in good agreement with the basic fracture data and size effect
data, and has been found convenient for programming. It is nowadays the main
concrete fracture model used in industry and commercial codes; e.g., code
DIANA, SBETA and ATENA.
Beginning with the mid 1980s, the stream of fracture and size effect studies
swelled to a torrent. Many researchers made significant contributions; to name
but a few: Petersson (1981), Hillerborg (1985), Elices et al. (1992), Planas et al.
(1994), Guinea et al. (1992), Carpinteri (1990) and Hu and Wittmann (1992).
An intriguing idea was injected in 1994 into fracture mechanics of concrete by
Carpinteri et al. (1994). They studied the effect of the invasive fractal character of
crack surface and the fractal character of micro-cracking in the FPZ. Although
significant doubts were raised with regard to proposals for a fractal explanation of
the size effects in structures, the fractal approach might be one useful way to
describe the effects of roughness and disorder on the fracture energy of concrete.
The purpose of this part is to survey the basic models for concrete fracture and
highlight some recent results.



2. Relative FPZ size and the paramount role of scale

Concrete as well as rock and many other materials requires a different kind of
fracture mechanics than metals. In both metal and concrete structures of normal
sizes, fracture mechanics is nonlinear due to the development of a sizable
nonlinear zone that develops at fracture front. But whereas, in ductile-brittle
materials, most of this zone involves hardening plasticity or perfect yielding, and
the FPZ, defined as the zone in which the material undergoes softening damage,
is quite small, in concrete and rock fracture the plastic flow is next to nonexistent
and the nonlinear zone is almost entirely filled by the FPZ (Fig. 1). Such
materials are now commonly called quasi-brittle.



Fig. 1: (a) Illustration of structural size effect in failure; (b-d) Relative sizes of fracture process
zone (F), nonlinear hardening zone (N) and linear zone (L).

As a second difference form metals important for concrete and geotechnical
structures, the length of the FPZ, which is equal or proportional to the so-called
characteristic length (or material length), l, may occupy a much larger portion of
the cross-section of the structure. Often the FPZ may encompass the whole
cross-section. The scale and size govern almost everything in fracture. In normal
concrete, typically the FPZ length m l 5 . 0 ~ ; in dam concrete with extra large
aggregates, m l 3 ~ , and in a jointed rock media, m l 50 ~ may be typical.
Depending on structure size D, understood as the dimension of the cross-section,
different theories are appropriate for analyzing failure. They may be
approximately delineated as follows:

For : 100 / > l D LEFM
For : 100 / 5 < s l D Nonlinear quasibrittle fracture mechanics
For : 5 / < l D Nonlocal damage, discrete element models, plasticity

Thus, for example, a massive rock slide in a 1000m tall mountain, consisting of
rock with joints spaced about 5m apart, may be analyzed by LEFM. On the other
hand, the fracture of a concrete arch or gravity dam 100m tall necessitates
quasibrittle fracture mechanics. The punching of a concrete slab 15cm thick can
conveniently be handled based on plastic limit analysis.
The basic criterion in fracture mechanics is that of energy release needed to
create the crack surface. According to the classical linear fracture mechanics,
N
o in fig.1a is then proportional to (size)
-1/2
, i.e., the plot of log
N
o versus log(size)
is a straight line of downward slope -1/2. (fig.1a). However, with the exception of
very large structures, this slope appears to be too steep in comparison with most
existing test data.
The reality seams to be a gradual transition from the horizontal straight line for
the strength criterion to the inclined straight line of slope -1/2 (fig.1a). the reason
for the deviation from linear fracture mechanics observed in concrete (as well as
rocks) consists in its heterogeneity of the material, causing that it behaves
nonlinearly within a relatively large zone adjacent to the fracture front, while the
linear fracture mechanics requires this zone to be small. In concrete, the fracture
process zone representing that part of the nonlinear zone in which the material
undergoes progressive microcracking manifested by strain-softening, is often
very large (fig.1d).


3. Cohesive crack and crack band models
3.1. Material characterization and cohesive crack model

Conceptually the simplest model to characterize the behavior of a finite-size FPZ
is the cohesive crack model (for concrete also known under the name fictitious
crack model). This model is simple enough to be understood even by someone
who has no knowledge of fracture mechanics. The basic hypothesis of the
cohesive crack model is that, for mode I fracture, the FPZ of a finite width can be
described by a fictitious line crack that transmits normal stress ) (x o and that this
stress is a function (monotonically decreasing) of a separation w (called also the
opening displacement, or opening width);

) (w f = o
(1)


(Fig. 2b, d and e). By definition, = ' =
t
f f ) 0 ( direct local tensile strength of
concrete (in ACI notation). The terminal point of the softening curve ) (w f is
denoted as
f
w ; 0 ) ( =
f
w f .


Fig. 2: Stress distributions and softening curves: (a,b) cohesive crack model for ductile-brittle
metals; (c,d) cohesive crack model for quasibrittle materials (concrete); (e,f) crack band model for
quasibrittle materials

The area under the entire softening stress-separation curve ) (w f represents the
total energy dissipated by fracture per unit area of the crack plane,
F
G , as the
crack faces are completely separated at a given point.
As a result, the two main parameters of the curve ) (w f are: the area under the
complete curve, representing the fracture energy
F
G , and (2) the initial slope of
the curve that may be characterized by the area under the initial tangent
representing the initial fracture energy
f
G (Fig. 2d).
In the light of the generally accepted bilinear approximation (Fig. 3a) of the
softening curve, the cohesive crack model is characterized by two fracture
energies:

= '
'
=
'
'
= =
0
0
0
2
0
0
2
)
) 0 (
(
2 2
, ) (
dw
df w f
G dw w f G
t
f F
o
o
o

(2)




Fig. 3: Bilinear softening stress-separation law: (a) Example of a softening law giving virtually the
same results; (b) A law lacking objectivity in general, acceptable only for a fixed crack path; (c)

For the scaling and size effect, it is important to realize that the fracture energy
and material strength imply, according to dimensional analysis, the existence of a
fracture characteristic length as a material property as shown by Irwin (1958). In
view of the bilinear approximation of softening curve, concrete possesses two
fracture characteristic lengths:

2 2
1
and
t
F
ch
t
f
f
EG
l
f
EG
l
'
=
'
=
(3)

The cohesive crack model has been widely regarded as a fundamental model
providing a yardstick for evaluating the soundness of all other models. This is not
quite true, though, for two main reasons:
(1) The cohesive crack model is strictly uniaxial model.
(2) In the cohesive crack model, a very tortuous crack with the adjacent zone of
microcracking and frictional slips is replaced by an ideal straight line crack, which
introduces some error.

3.2. Crack band and nonlocal models

From the viewpoint of computational effectiveness, the line crack model appears
to be disadvantageous. When the crack extends through a certain node, the
node must be split into two nodes, increasingly the total number of nodes and
changing the topological connectivity of the mesh. These difficulties are avoided
by the crack band model. The crack band model has been the model most widely
used in practice for analyzing the distributed cracking and fracture of concrete,
and has been incorporated in a number of commercial finite element codes.
Aside form programming convenience, its advantage over the cohesive crack
models is that it can take into account triaxial stresses in FPZ, particularly the
normal and shear stresses acting in the directions parallel to the crack plane.
Still another advantage is the fact that with the crack band model one can treat
the case when principal stress directions in the fracture process zone rotate
during the progressive fracture formation, i.e., during the strain-softening.


Fig. 4: (a) Actual crack morphology; (b) Actual stresses and then smoothing; (c) Line crack model;
(d) Crack band model used here

Recently, various measurements are being made to observe the formation of
microcracks at the fracture front. From these observations it seems that the
larger microcracks that can be seen are not spread over a band of a large width
but are concentrated essentially on a line. However, the long along which the
microcracks are scattered is not straight but is highly tortuous (fig. 4), deviating to
each side of the straight line extension by a distance equal to about the
aggregate size, as the crack is trying to pass around the harder aggregate pieces.
In the equivalent, smoothed macroscopic continuum which is implied in structural
analysis, the scatter in the locations of visible microcracks relative to a straight
line is characterized by a microcrack band better than by a straight row of
microcracks.


Fig. 5: (a-d) Stress-strain diagrams for fracture process zone; (e) Zig-zag crack band; (f-g) Stress
distribution in fracture process zone

We will work with a system of Cartesian coordinate axes and will treat concrete
as an isotropic elastic material characterized by Youngs modulus E and Poisson
ratio v . Consider that a system of densely and uniformly distributed
discontinuous microcracks normal to axis z develops in the material while the
stresses are kept constant. This must lead to an increase of strain
z
c , but the
effect of this on strains
y x
c c , parallel to the microcracks should be nil.
Then, assuming that
z y x
o o o , , are the principal stresses and
z y x
c c c , , are the
principal strains, we have:

f z
y
x
z
y
x
E
c o
o
o
v v
v v
v v
c
c
c
0
0
1
1
1
1

(4)

Where
f
c if the fracture strain. i.e., the additional strain caused by the opening of
the microcracks, E = Youngs elastic modulus of concrete, and v = its poissons
ratio.
As generally accepted, the front of an advancing crack band (microcrack zone),
called also FPZ, has a certain characteristic width
c
w (fig.4). For plain concrete,
we may consider
c
w as a material constant that can be determined by
experiment. We expect that
c
w is several-times the maximum aggregate size.
Now, the meaning of
f
c is the average over the FPZ of the deformation due to
microcracking, precisely
c f f
w / o c = where

=
i
i
f f
o o = sum of the openings of
individual microcracks intersecting axis z (fig.4).
It is probably reasonable for us to assume that
f
c is a function of
z
o , i.e.,
). (
z f
f o c = Obviously, it must be a monotonically decreasing function (fig 5a,b).
The simplest choice is a linear function, i.e. :

) (
1
) (
z t
f
z f
f
C
f o o c ' = =
(5)

Substituting equation (5) into equation (4) we obtain, for the strain-softening
range:



0
1
1 1
1 1 1
0
0
Sym c o
o
o
v
v v
c
c
c
z
y
x
t z
y
x
E
E E
E E E

(6)

In which

0
1 1 1
s =
f t
C E E

(7)

t
E is the tangent softening modulus of the declining (strain-softening) segment of
the uniaxial stress-strain diagram in the z direction (fig.5).; and
0
c is the strain at
the end of strain-softening (fig.4), at which the microcracks coalesce into a
continuous crack and
z
o vanishes;
f t
C f /
0
' = c ; the square matrix in equation (6)
is the tangent compliance matrix .

Fracture Characteristics in crack band model
The fracture energy,
f
G is the energy consumed in the formation and opening of
all microcracks per unit area of plane (x,y). Thus:

' =
'
= = =
0 2
2
0
2 2
1
t z
f
c
f
t
c f f z c f
w
C
f
w C d w G
o
c c o
(8)

In which
c
w is the effective width of the fracture process zone (or crack band)
over which the microcracks are assumed to be uniformly spread, and
t
f ' is the
direct tensile strength. If
f
G ,
t
f ' and
c
w are known from measurements, then the
basic parameters of our stress-strain relation may be calculated as:

c t
f
f
t
f
c t
f
w f
G
C
f
G
w f
C
'
=
'
=
'
=
2
,
2
0
2
c
(9)

It is interesting to note that equation (7) allows us to link the area W under the
complete uniaxial stress-strain diagram with the fracture energy. Substituting for
f
C form equation (8), equation (7) becomes
c t f t
w f G E E
2 1 1
/ 2 ' =

, from which:

c c p t p t
c t
t
f
Ww w f f
w f
E E
G
= ' + ' =
' =
)] ( [
2
1
)
1 1
(
2
1
0
2
c c c

(10)

Where

= c o d W
z
. This conclusion, however, holds only if no plastic
deformations accompany tensile microcracks as we tacitly assumed.

Computational aspects
In the general case of a fracture that is not parallel to the mesh lines, the crack
band has a zig-zag shape (fig. 5e). we must however keep in mind that this is
just a numerical model for a smooth crack band, and we must therefore decide
the value
c
w of the width of a smooth band which corresponds to our zig-zag
band. Restricting attention to square meshes, it seems most reasonable to
assume that = =
c c
l w characteristic length of an element which defines as
A l
c
= (or V ) where A and V are element area and volume, respectively.
Thus, it should not matter when an element-wide crack band of width
c
l is
considered instead of the actual width
c
w of the fracture process zone, provided
that the correct value of the fracture energy is preserved. So we may adjust the
softening modulus
t
E as long as
f
G remains the same. Replacing
c
w by
c
l , we
see that the following adjustments are needed when
c
l is chosen larger than
c
w :

)
2
1
(
2
,
2
2
0
2
c t
f
t
c t
f
f
c t
f
l f
G
E
E
l f
G
G
l f
C
'
=
'
=
'
= c

(11)

Since the tangent strain-softening modulus
t
E must be negative, the adjustment
in equation (11) can be used only if:

2
2
t
f
c
f
E G
l
'
<


In practice, the finite element should be less than about of this value.

S-ar putea să vă placă și