Professional Documents
Culture Documents
Effects of The Early Age Thermal Behavior On The Long Term Damage Risks of Massive Structures
Effects of The Early Age Thermal Behavior On The Long Term Damage Risks of Massive Structures
1. Introduction
The temperature evolutions are obtained by solving the heat equation. This
equation includes the heat release due to the hydration reaction (through a heat
source term):
CT kT L [1]
where L is the total heat release [J.m ], k is the thermal conductivity [W.m .K-1]
-3 -1
In the literature, (for example (Boussa, 2000), (Azenha and al., 2009), (Jeon and
al., 2008), (Lee and al., 2009)), a value of about 10W.m-2.K-1 for the coefficient of
exchange by convection is generally used. Besides, to simplify the model, this
parameter includes the exchange by natural or forced convection but also the
radiation one ((Buffo-Lacarrière, 2007) for example). The calculation of the
convective coefficient can be done through the number of Nusselt (Nu):
hl
Nu c [4]
kf
where lc is the characteristic length [m] of the convective flux and kf is the thermal
conductivity of the fluid [W.m-1.K-1].
Although, the Nusselt number can also be calculated from the following
equations:
Nu 0,13Pr Gr
1/ 3
For natural convection: [5]
g T lc,Gr
Gr [7]
2
where g is the gravitational acceleration [m.s-2], is the thermal dilatation
coefficient of the fluid [K-1], lc,Gr is the characteristic length of the natural
convection [m], is the fluid kinematic viscosity [m².s-1].
Pr [8]
4 Revue. Volume X – n° x/année
V l
c, Re
Re [9]
where lc,Re is the characteristic length of the forced convective flux [m], V is the fluid
velocity [m.s-1].
Because the temperature variation remains little compared to the radiation heat
transfer, radiation heat transfer can be linerarized (Hernot and al., 1984) in order to
be integrated to the global convective coefficient:
1/ 2 1 / 3
V lc, Re kf
3 [12]
h 0,664 4 Tmoy
For forced convection:
l
c, Re
For example, for a massive wall (thickness: 1.2m; height: 2m; length: 20m) and
with the assumption that the wind direction is parallel to the wall, figure 1 shows the
evolution of the global convective coefficient. One can see on this figure that for a
wind velocity equal to 0, usual values are included between our maximal and
minimal value (the evolution begins with 5.3W.m-2.K-1 and reaches 13.3 W.m-2.K-1
for a high difference between the surface and the external temperature). The
framework could be integrated in the convective coefficient (by the use of an
equivalent conduction coefficient for instance).
Massive structures long term damage risks: early age thermal behaviour effect 5
30,00
25,00
Text
20,00
Twall
h 15,00
(W/m²K)
10,00
5,00
25,00-30,00 Wind
20,00-25,00 0,00
0
10
20
15,00-20,00
30
40 0 Wind
50
60
70
10,00-15,00
80
velocity
5,00-10,00 (Km/h)
0,00-5,00 Text-Twall (°C)
X
Z
50
#2
10
45 10
#1
Raft fondation
Fondation
40
T1 h variable
35 T2 h variable
T3 h variable
30
T5 h variable
25 T6 h variable
T1 h constant
20 T2 h constant
0 20 40 60 80 100 120 T3 h constant
time (h) T5 h constant
T6 h constant
Investigating the maximal temperature reached at early age can be a good way to
study the massive structure sensibility to the risks of delayed ettringite formation
(DEF). Indeed, even though DEF is a long term deterioration (Taylor, 1997),
(Lawrence, 1998) the dissolution of ettringite composed phases is a necessary
conditions for the appearance of DEF. It may appear if at early age the maximal
temperature exceeds 65-70°C. Of course, this condition is not sufficient and DEF
needs other factors to occur (Collepardi, 2003) (LCPC recommendation, 2007).
Nevertheless, in this paper, a study of maximal temperature reached in a massive
wall (thickness: 1.2m; height: 2m; length: 20m), cast on a massive slab, is proposed
(this case corresponds to the first lift of a nuclear containment construction).
In this study, only the external conditions (wind velocity and external
temperature) are source of variability but for an application to a real case, materials
properties variability must be taken into account (total heat release, thermal concrete
conduction,…). Besides, the initial concrete temperature has a significant effect on
the maximal temperature but to avoid this third parameter, we used a relation
between the external temperature (Text) and the initial concrete one (Tini), proposed
by (Torrenti and al., 2009). The initial concrete temperature is linked to the external
temperature because aggregates are generally kept in stock outside but it often is
higher than external temperature because cement is generally at a higher
temperature. Figure 4 shows the comparison between external temperature
measurement and initial concrete temperature measurement on two building sites.
Massive structures long term damage risks: early age thermal behaviour effect 7
The results of this study are presented on figure 5. One can see on this figure a slight
effect of wind velocity on maximal temperature. This can be explained by the fact
that, as the thermal concrete conductivity is rather low, the thermal exchange
conditions at the core are close to adiabatic conditions. On the contrary, the
evolution of the maximal temperature due to the external temperature is quasi-linear.
This can be explain by the fact that the initial temperature is a linear function of the
external temperature.
60-62
60
55-60
55
Tmax (°C)
50-55
50
45 45-50
40 40-45
35
35-40
40
20
30
15
10
20
0
0
2.4 Effect of wind velocity and external temperature on the risks of cracking due
to thermal gradient
29
28-29
28
27-28
27 26-27
26 25-26
Tsurface - Text
(°C) 24-25
25
23-24
24
22-23
23 21-22
22
21 20
40
30
10
20
10
Figure 6. Temperature difference evolution between the core and the surface (Tsurface
–Text) of a massive wall (thickness 1.2m): effect of the wind velocity and of the
external temperature
Massive structures long term damage risks: early age thermal behaviour effect 9
2.5 Effect of the evolution of the concrete thermal properties evolutions at early
age
k k 1,33 0,33 [14]
10 Revue. Volume X – n° x/année
where k∞ stands for the value of thermal conductivity for hardened concrete [W.m-
1 -1
.K ] (ξ= ξ∞).
In the IPACS project (Improved Production of Advanced Concrete Structures),
Lura and al. (Lura and al. 2001), proposed a thermal capacity decrease linked to the
hydration degree:
cbindwwcement Cwater
C C0 [15]
concrete
where C0 is the thermal capacity calculated from the concrete mix, is the hydration
degree [-], cbindw is a coefficient which takes into account how chemically and
physically bond water influences the thermal capacity (equal to 0.2), Wcement is the
cement content of the mix [Kg.m-3], Cwater is the thermal capacity of the water,
concrete is the concrete density .
2.5.3 Effect of temperature and hydration degree on thermal conductivity and
thermal capacity: simulation results
Simulations have been performed on the same geometry than the simulations
studying the effect of a variable convective coefficient (figure 3).
Figure 7 displays the results obtained with different thermal conductivity
evolution. The evolution represented with white dots (square, circle and triangle) are
the results of the simulations with a constant thermal conductivity (equal to 2.5
W.m-1.°C-1). The upper and the lower values, represented by the vertical bar,
correspond to simulations with a constant thermal conductivity (equal 2 and 3 W.m-
1
.°C-1 respectively). The black dots (square, circle and triangle) correspond to the
evolution which takes into account effect of temperature and hydration degree
(eq.13 and 14). For this simulation, the final thermal conductivity used is equal to
2.15 in order to obtain a mean value of 2.5 at ∞=0.42 (half of the hydration
reaction). One can see that temperature at the core of the concrete element (T3) is
slightly decreased by the use of a variable thermal conductivity.
Massive structures long term damage risks: early age thermal behaviour effect 11
65
T1_k=2.5
60 T1_k=f(T,ξ)
55 T2_k=2.5
T2_k=f(T,ξ)
50
T3_k=2.5
Temperature (°C)
45 T3_k=f(T,ξ)
40
35
30
25
20
0 20 40 60 80 100 120
Time (hours)
The study of the thermal capacity variation is more complex because this one is
also used in the determination of the total heat release and the chemical affinity. The
concrete thermal capacity is used in the analysis of semi-adiabatic test which is the
base of the determination of the adiabatic temperature evolution (Ulm et al. 98).
In this test, the heat release q is calculated from the concrete temperature
evolution Tconcrete and the external temperature Text :
qt sa Cconcrete Tad t sa Tad t sa 0
t sa [16]
a b t t dt
Ctot Tconcrete t sa Tconcrete t sa 0
0
where Cconcrete is the concrete thermal capacity [J/°C], Ctot is the total thermal
capacity (concrete + calorimeter), a [W] & b [W.°C-1] are the heat losses coefficient
of the calorimeter, tsa [s] is the real time of the semi-adiabatic test, = Tconcrete - Text,
and Tad is the theoretical adiabatic temperature.
Table 1. Value of adiabatic temperature obtained for different thermal capacity
value
Table 1 gives the value of adiabatic temperature and figure 8 represents the
affinity evolution obtained for different thermal capacity values. In this calculation,
the effect of the hydration degree variation on the thermal capacity is not taken into
account because its evolution is calculated from the results of the semi-adiabatic
test.
A(ξ) (C=900 J/(Kg.K))
1600
A(ξ) (C=1050 J/(Kg.K))
1400
A(ξ) (C=1200 J/(Kg.K))
Chemical affinity (1/s)
400
200
0
0,00 0,10 0,20 0,30 0,40 0,50 0,60 0,70 0,80 0,90
Figure 9 displays the effect of thermal capacity. The evolutions represented with
white dots (square, circle and triangle) are the results of simulations with a constant
thermal capacity (equal to 1050J.Kg-1.K-1). The upper and the lower value,
represented by a vertical bar, correspond to simulations with constant thermal
capacity (values of 900 and 1200 J.Kg-1.K-1 respectively). The black dots (square,
circle and triangle) correspond to the evolution taking into account the effects of
temperature and hydration degree (eq. 13 and 15). For this simulation, the initial
thermal capacity used is equal to 1111J.Kg-1.K-1 in order to obtain a value of
1050J.Kg-1.K-1 at half of the hydration reaction (∞=0.84). A decrease of about 8%
on the temperature at the core of the concrete element (T3) by the use of a variable
thermal capacity is obtained. Besides, one can see that the effect of a variation of the
thermal capacity (from 900 to 1200J.Kg-1.K-1) has a significant non-linear influence
on the maximal reached temperature.
Massive structures long term damage risks: early age thermal behaviour effect 13
70
T1_C=1050 J.Kg-1.K-1
65 T1_C=f(T,ξ)
T2_C=1050J.Kg-1.K-1
60
T2_C=f(T,ξ)
55 T3_C=1050J.Kg-1.K-1
Temperature (°C)
T3_C=f(T,ξ)
50
45
40
35
30
25
20
0 20 40 60 80 100 120
Time (hours)
50
Temerature (°C)
45
40
35
30
25
20
0 20 40 60 80 100 120
Time (h)
Figure 10. Temperature evolution of a massive wall (thickness 1.2) at the surface or
at the core: effect of the variation of thermal properties with respect to hydration
degree and temperature
The Young modulus E increases due to hydration as follows (De Schutter 1999)
(Stefan 2010):
0
E E [18]
with 0
ε ijau 0 ij [19]
ε ijth (T T0 ) ij [20]
kbci i
bci bci bci 1bc i [21]
kbc
i
kbc
where bc
i
is the characteristic time (constant), kbc
i
is the stiffness of the spring
(increasing with the hydration degree), is the apparent stress.
The stiffness parameter for each unit is calculated with the following equation (De
Schutter 1999):
kbci kbci _
0.473
0.62 [22]
2.081 1.608
The effect of temperature on creep must be taken into account. Experimental results
highlight that temperature increases creep at early-age by a factor of 2 or 3
(Arthanari 1967). On the one hand, temperature evolution increases the rate of
hydration degree, and therefore reduces creep strains at higher temperature. Indeed,
16 Revue. Volume X – n° x/année
if temperature increases, the hydration degree (eq.2) and, therefore, each spring
stiffness increase (eq. 22). The result is a decrease of the creep strain rate with
respect to temperature. This is in contradiction with experimental evidences. On the
other hand, temperature increases creep strains due to the two following factors
(Hauggard and al. 1999):
At constant temperature, creep strains rate increases. This is due to the
decrease of water viscosity with temperature;
Transient temperature history also increases creep strains. The obtained
deformation is called transient thermal creep or load induced thermal
strains. Such kind of strain is observed at very high temperature (above
100°C) but exists at lower ones (Bažant and al 1997). Bažant and al
(Bažant and al 1997) suggest that this strain corresponds to drying creep.
The effect of constant temperature on creep strain is taken into account by
modifying the spring stiffness and the dashpot viscosity of Kelvin-Voigt units as
follows:
Eac 1 1
kbci , T kbci , T0 e R T T0
[24]
Eac 1 1
bci , T bci , T0 e R T T0
[25]
where Eac is the creep activation energy and T 0 = 293 K. In this way, the
characteristic time is kept independent of the temperature.
Hauggard and al.(Hauggard and al. 1999) have modeled transient thermal creep by
using the model developed by Bažant and al. (Bažant and al 1997). Here, a simpler
model is used which is similar to the one used to predict thermal transient creep of
concrete at high temperatures (Thelandersson, 1987). Thermal transient creep strain
ttc reads:
ttc T ~ [26]
removed and the wall is free to expand or to shrink. Nevertheless, the temperature
fields have been calculated on the complete geometry.
Four simulations for four cases were performed:
1. The thermal capacity and the thermal conduction are constant values
2. The thermal capacity is function of hydration degree and temperature but
the thermal conduction is a constant value
3. The thermal capacity is a constant value but the thermal conduction is
function of hydration degree and temperature
4. The thermal capacity and the thermal conduction are function of hydration
degree and temperature
The other thermal parameters and all the mechanical parameters are the same for
each simulation.
For example, figure 11 displays the stress fields for the case N°3 when temperature
at the core reaches the peak.
Figure 11. Example of stress fields (MPa) generated by self restraint for a wall of
1.2m width.
Case 1 2 3
Table 2 summarizes the stress variability obtained for each direction and for each
case with respect to the case which is closer to reality (the thermal capacity and the
thermal conductivity depend on the hydration degree and the temperature).
The results show that, not taking into account the evolution of thermal properties
can lead to an error on the maximal stress calculated of about 30%. Besides, two
conclusions can be drawn. Firstly, although in this configuration the range of
generated stress by self restraint is low, thermal properties variability cannot be
neglected in an accurate study of cracking due to self restraint. Secondly, even
though it is conservative, constant values for the thermal properties induce a
significant overestimation of the stresses.
4. Conclusions
The thermal early age behaviour influence on the concrete damage mechanisms
can be declined under two main axes:
The sensibility to DEF risks
The sensibility to concrete skin cracking due to thermal gradient
In this paper, the effects of environmental conditions during the casting and the
hydration reaction were studied. The results show that, if the wind velocity is low,
the use of a constant convective coefficient gives temperature field with sufficient
accuracy. Besides, it has been shown that the wind velocity effect on the DEF risks
is weak but non-negligible on the cracking by thermal gradient risks. As the effects
of external temperature are the opposite, there are no optimal casting conditions.
A parametrical study highlights the effect of the thermal properties (concrete
thermal conductivity and thermal heat capacity) variation (in function of the
temperature and the hydration degree) on the maximal temperature reached and on
the maximal temperature difference between the core and the surface
(overestimation of 10.4% on the maximal temperature and of 14.4% on the
temperature difference). Moreover, with the use of constant thermal parameter, the
value of the latter could have a significant and non-linear influence on the reached
maximal temperature.
Finally, a mechanical study with a visco-elastic model show that the
overestimation of the temperature difference between the core and the surface with
the use of constant parameter leads to an overestimation of about 30% on the
maximal generated stress. As the overestimation is significant, thermal properties
evolution cannot be neglected for the study of massive structure self-restraint.
5. Bibliographie
(Azenha and al., 2009) Azenha M., Faria R., Ferreira D. (2009), Identification of
early-age concrete temperatures and strains: Monitoring and numerical
simulation. Vol. 31. Cement & Concrete Composites, p. 369-378.
Massive structures long term damage risks: early age thermal behaviour effect 19