128
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
2.3 Compressive test results
The experimental compressive creep results of [8] for four loading levels (25%, 35%, 50%
and 65% of the compressive strength) were adopted. The creep law parameters were
calibrated for the lowest loading level whereas the mechanical properties were calibrated with
ordinary cement paste values [7] and to obtain no damaged element after the instantaneous
loading at 25% of the compressive strength.
A comparison between the experimental results and the mesoscopic approach with and
without damage are presented in Table 1 and Figure 2. The figure and the table highlight that
the simulation of creep evolution with damage are closer to the experimental ones than the
simulation curves without damage. Assuming that creep of concrete occurs without damage
means that there are no micro-cracks at the interface between cement paste and aggregate.
Consequently, there is no additional creep strain that occurs normally due to these micro-
cracks. The mesoscopic approach reveals the nonlinearity between the creep strain and the
applied load. It could be seen, for example, that for the highest loading level, the percentage
of nonlinearity due to the incompatible strains and the associated damage is about 66%.
Nevertheless, if an accurate prediction can be obtained with 2D model for tension test, for
compressive test, 3D modelling is required.
By comparing the field of damage for two loading levels 25% and 65%, one can remark that
the damage amount is higher for 65% than that for 25% as expected. That is related to the
tensile stresses that develop at the cement paste - aggregates interface and lead to the cracking
development. As the creep strain of cement paste is directly related to the loading level, the
damage at the interface between cement paste and aggregates increases.
Figure 2: Compressive creep strain evolution: comparison between experimental data and
mesoscopic approach with and without damage
0
50
100
150
200
250
-3500
-3000
-2500
-2000
-1500
-1000
-500
0
500
Time (days)
St
ra
in
(
μ
m/
m)
sim. 25%
sim. 35%
sim. 50%
sim. 65%
sim. without damage 65%
Exp. 25%
Exp. 35%
Exp. 50%
Exp. 65%
129
International RILEM Conference on Materials, Systems and Structures in Civil Engineering
Conference segment on Service Life of Cement-Based Materials and Structures
22-24 August 2016, Technical University of Denmark, Lyngby, Denmark
Table 1: Comparison of the experimental compressive creep results with those of mesoscopic
approach with and without damage (Nonlinearity% = 100 (
wd
-
wod
) / (
exp
-
wod
)
Loading level
Exp. Results
Sim. Results
Explained
Nonlinearity
MPa
/
25%
(
exp
) 10
-6
at
210days
With damage
(
wd
) 10
-6
Without damage
(
wod
) 10
-6
%
10.5 1.4 1153
1026
594
77.2
15 2 2108
1719 836
69.4
19.5 2.6 3183
2476
1087 66.3
2.4 Numerical mechanical behaviour after creep loading in compression
Fichant
[
12] developed two damage models (isotropic model and orthotropic model) to solve
problems with more complex loadings. In these models, the damage development is coupled
with the cracking energy. When the radial loading is applied, the isotropic model is adequate
while for the disproportionate loading, the orthotropic model is adopted. In both models, a
coupling could be done with plasticity and unilateral effects.
The damage affects the elastic part of the stress-strain relation behavior [12]:
kl
ijkl
ij
C
(3)
Where
ijkl
C
is the stiffness of the damaged material whereas
ij
and kl
ijkl
C
represent the
components of the stresses tensors and elastic strain tensor.
The equivalent strain is calculated from the elastic strain
e
and the damage evolution law is
expressed as follows:
))
(
(
exp
eq
0
d
t
eq
0
d
B
1
D
when
0
D
(4)
where
0
d
is the threshold in tension. The parameter
t
B
is calculated as a function of cracking
energy
f
G
and of the element size
h
with the following equation:
f
t
t
G
f
h
B
/
(5)
where
t
f
is the tensile fracture stress of the material. This regularised technique based on the
proposition of [13] allows to avoid strong mesh dependency.
In this study, the isotropic model is used without plasticity and unilateral effect. Moreover,
damage in compression is also unconsidered. Indeed, at the mesoscale,
the rupture is assumed
to be only due to extension. The geometric representation of the constituents and the contrast
between material properties is assumed the source of complex behavior observed at the
macroscopic scale.
Figure 3 represents the constitutive law of compressive behaviour after creep in compressive
had taken place. It reveals that the maximum strength and in the Young modulus are lower for
the concrete specimens that were under a higher creep loading and all are lower than the one