Professional Documents
Culture Documents
FraMCoS-8
J.G.M. Van Mier, G. Ruiz, C. Andrade, R.C. Yu and X.X. Zhang (Eds)
* * †
JAN CERVENKA , LIBOR JENDELE AND VIT SMILAUER
*
Cervenka Consulting s.r.o
Na Hrebenkach 2667/55, 150 00, Praha 5, Czech Republic
e-mail: jan.cervenka@cervenka.cz, www.cervenka.cz
†
Czech Technical University, Prague (CTU)
Thakurova 7, 160 00, Praha 5, Czech Republic
e-mail: smilauer@post.cz, www.fsv.cvut.cz
Abstract: A suitable control of cracking of young concrete often determines the durability and the
reliability of concrete structures. A hydro-thermo-mechanical approach is presented for analysis of
reinforced concrete structures applicable in engineering practice. The multi-physics problem is
solved by a staggered analysis that consists of two steps. The moisture and heat transport analysis is
performed and the results are time dependent moisture and temperature fields in all material points.
These resulting fields are then used in a stress analysis to calculate stresses, crack patterns and crack
width. The whole model is implemented in a finite element software [1], which is used in the
validation as well as in a practical application example.
1
J. Cervenka, L. Jendele and V. Smilauer
2
J. Cervenka, L. Jendele and V. Smilauer
3
J. Cervenka, L. Jendele and V. Smilauer
4
J. Cervenka, L. Jendele and V. Smilauer
2
computed by the linear creep model described ρ ρ ξ
above. F = 1.5 + m
p
3P r (θ , e) + −c = 0
fc 6 fc 3 fc
The short-term concrete mechanical model
(25)
follows the original theory in [15]. The
material model formulation is based on the where
strain decomposition into elastic ε ije , plastic
f c2 − ft 2 e
ε ijp and fracturing ε ijf components [16]. m=3
fc ft e + 1
4(1 − e 2 ) cos 2 θ + (2e − 1) 2
ε ij = ε ije + ε ijp + ε ijf (22) r (θ , e) = 1
2(1 − e 2 ) cos θ + (2e − 1) 4(1 − e 2 ) cos 2 θ + 5e 2 − 4e 2
The new stress state is then computed by (26)
the formula: In the above equations, (ξ , ρ ,θ ) are Heigh-
σ ijn = σ ijn −1 + Eijkl (∆ε kl − ∆ε klp − ∆ε klf ) (23) Vestergaard coordinates, and f c and ft are
Tensile behaviour of concrete is modelled compressive strength and tensile strength
by non-linear fracture mechanics with a simple respectively. Parameter e ∈ 0.5,10. defines the
Rankine-based criterion. roundness of the failure surface.
The surface evolves during the
Fi f = σ i′t − ft ′i ( wi′) ≤ 0 (24) yielding/crushing process by the
A smeared crack concept is adopted with hardening/softening laws based on equivalent
the following parameters: tensile strength f t , plastic strain defined as
shape of the stress-crack opening curve f t ( w) ∆ε eqp = min(∆ε ip ) (27)
and fracture energy GF . It is assumed that
Hardening ε eqp ∈ −ε cp ; 0 :
strains and stresses are converted into the
material directions, which in a case of rotated 2
crack model correspond to the principal stress ε cp − ε eqp
f c (ε ) = f co + ( f c − f co )
p
1− (28)
directions, and in a case of a fixed crack model eq εp
c
to the principal directions at the onset of
cracking. Therefore, σ i′t identifies the trial Softening ε eqp ∈ −∞; −ε cp :
stress and f t′i tensile strength in the material 2
w
direction i . The prime symbol denotes c = 1 − c , wc ∈ − wd ; 0
quantities in the material directions. This wd
approach is combined with the crack band c = 0, wc ∈ (−∞; wd ) (29)
method of Bažant and Oh [17]. In this
formulation, the cracking strain is related to wc = (ε eqp − ε cp ) Lc
the element size. Consequently, the softening When concrete crushing enters into the
law in terms of strains for the smeared model softening regime, an analogous approach to
is calculated for each element individually, the crack band model is used also for the
while the crack-opening law is preserved. The localization in compression within the
model uses an exponential softening law of
crushing band Lc . A direct return-mapping
Hordijk [18].
The compressive behaviour is modelled algorithm is used to solve the predictor-
using a plasticity-based model with failure corrector equation of the plasticity model.
surface defined by the three-parameter F p (σ ijt − σ ijp ) = F p (σ ijt − ∆λlij ) = 0 (30)
criterion form [19]
5
J. Cervenka, L. Jendele and V. Smilauer
The plastic stress σ ijp is a product of plastic points C,D located on the beam axis 2.5 m
apart.
multiplier ∆λ and the return direction lij ,
In the first step, the adiabatic experiment
which is defined as follows was used to validate the hydration model. The
∂G p (σ klt )
model predicts correctly the temperature rise
lij = Eijkl (31) by calibrating only the length of the dormant
∂σ kl period. Based on these results the four-
parametric affinity model was calibrated [8].
∂G p (σ ijt ) Parameters of the affinity model are B1=1.8 h-
∆ε ijp = ∆λ (32)
∂σ ij 1
, B2=1.0e-5, η=7.0, DoH∞ = 0.90. Figure 4
demonstrates that both models approximated
The plastic potential G p is given by reasonably well experimental data, especially
1 at younger ages. It reveals clear negative slope
G p (σ ij ) = β I1 + 2 J 2 (33) documenting the energy loss in the adiabatic
3 experiment.
where β determines the return direction. If
β < 0 , material is being compacted during
crushing, if β = 0 , the material volume is
preserved, and if β > 0 , the material is
dilating.
A special iterative algorithm [15] analogous to
multi-surface plasticity is used to solve the
plastic and fracture models such that the final
stress tensors in both models are identical.
The presented model is aimed to calculate
primarily reinforced concrete structures. In
such particular case, two material laws are
used; the above described model for adjacent Figure 1: A validated massive beam RG8 with
restrained shrinkage (image and data from the database
concrete and a 1D model for reinforcement. Cheops).
The latter model can be any time-independent
model, such a multi-linear 1D model, or The same numerical model was used for the
elasto-plastic model with hardening etc. heat/moisture transport and mechanical
analysis (see Figure 3). Note that the head
4 VALIDATION sizes were reduced and the struts are directly
A validation of the presented model shown next to the central beam surfaces.
using the example from ConCrack benchmark
experiment RG8. For further information see
http://www.concrack.org/. Figure 1 shows the
overall geometry of the beam RG8 with two
massive heads on both sides and two
restraining steel struts. The central beam part
has dimensions 0.5x0.8x5.1 m and is
reinforced. The benchmark experiment
provides data on cement mineral composition,
cement fineness, concrete composition,
Figure 2: Our model showing the reinforcement and
reinforcement, geometry, external temperature,
material assignment.
internal temperature in three points, adiabatic
concrete temperature and displacements of two
6
J. Cervenka, L. Jendele and V. Smilauer
7
J. Cervenka, L. Jendele and V. Smilauer
12 APPLICATION EXAMPLE
The presented model was applied in a
practical engineering problem of cracking in
newly built massive concrete columns. The
analyzed columns support the boiler-room of a
newly built 660MW coal power plant (see
Figure 8). Altogether 4 columns with
dimensions 5x5x4.57 m support the boiler
chamber with total weight of about 672 MN.
Figure 5: Validation of relative displacements of points Approximately 1 year after the construction of
C-D with mutual distance 2.5 m on the beam axis. The the massive columns, vertical cracks up to
coefficient ks shows stiffness reduction of the 0.5 mm were detected in the middle of the
struts.
columns (see Figure 9) . It was assumed that
the hydration heat, shrinkage and insufficient
reinforcement (Figure 10) might have been the
main reason for this cracking. The presented
model was used to confirm this assumption
and to evaluate the effect of this initial damage
on the strength and reliability of these critical
load bearing elements.
8
J. Cervenka, L. Jendele and V. Smilauer
9
J. Cervenka, L. Jendele and V. Smilauer
REFERENCES
Figure 13: Comparison of load-displacement response
for the analysis without (i.e. static) and with temperature
[1] V. Cervenka, J. Cervenka, and L.
and shrinkage effects.
Jendele, "Atena Program
The load carrying capacity of the columns is Documentation, Part 1-7", Prague:
shown in Figure 13. It compares the load- Cervenka Consl.,2000-2012.
displacement response of a pure static analysis [2] M. Nilsson, "Thermal Cracking of
with the analysis based on the presented Young Concrete. Partial Coefficients,
hydro-thermo-mechanical model. The result of Restraint Effects and Influence of
the new model is depicted by a solid line. Casting Joints", Thesis, Lulea, Sweden,
Initially, there is no load, but the vertical Lulea Univeristy of Technology, ,
deformation increases due to the thermal 2000.
effects, which is followed by a decrease due to [3] A.M. Neville, "Properties of Concrete":
shrinkage. After that the load is increased up John Wiley & Sons, Inc.,1997.
to 40% of the design load to simulate the [4] D. Gawin, F. Pesavento, and B.A.
situation after 720 days. This load level is kept Schrefler, "Hygro-thermo-
constant for 540 days while the creep and chemomechanical Modelling of
shrinkage further decreases the deformation. Concrete at Early Ages and Beyond.
This is represented by the short horizontal line. Part I: Hydration and Hygro-thermal
After that the column is loaded up to failure. Phenomena", Int. Journal for
This response can be contrasted by the pure Numerical Methods in Eng., 67(3),
static analysis depicted by the dashed line, 299-331, 2006.
which shows a smooth response up to failure. [5] R. Faria, M. Azenha, and J.A.
The analysis confirmed that the columns have Figueiras, "Modelling of Concrete at
a sufficient load carrying capacity, i.e. about Early Ages: Application to an
200% of the design load, but their strength was Externally Restrained Slab", Cement
significantly reduced by almost 30%. and Concrete Composites, 28(6), 572-
578, 2006.
12 CONCLUSIONS [6] M. Cervera, J. Oliver, and T. Prato,
"Thermo-chemo-mechanical Model for
The combined thermo-hydro-mechanical Concrete. I: Hydration and Aging",
model was presented suitable for modelling Jour. Eng. Mech. ASCE, 125(9),
the behaviour of young concrete structures. 1018-1027, 1999.
The behaviour of the model was demonstrated [7] C. Hellmich, H.A. Mang, and F.J. Ulm,
on a hydrating concrete beam with "Hybrid Method for Quantification of
experimental results. The developed model Stress States in Shotcrete Tunnel
was successfully applied to a practical
10
J. Cervenka, L. Jendele and V. Smilauer
11