7.1 Introduction
For quite a number of reasons, the behaviour of hardening concrete has gained more and more
interest in recent years. The most important reasons are:
For all these cases, emphasis is on the hardening process and the associated phenomena such as
temperature, strength and stress development and finally, cracking. Besides the presence of
thermal stresses, the use of low water-cement ratio concretes substantially contributes to the
development of stresses due to autogenous shrinkage (see section 7.6.3.).
material. Without the help of adequate numerical tools, quantification of this behaviour is
impossible.
Factors that play an important role for the quantification of the stress development and the
probability of cracking in early-age concrete are:
The factors mentioned, i.e. properties, show a time dependency. It is better to say that there exists
a relationship with the development (progress) of the hydration or hardening process. An accurate
description of the hydration process is therefore indispensable for achieving a consistent
description of the behaviour of hardening concrete.
The hardening process of concrete is the result of a chemical-physical reaction of cement and
water. This is an exothermic reaction, which is a reaction during which heat is liberated. Due to
this liberated heat, the temperature of the concrete rises (Fig. 7.1a) and the concrete expands. The
strain increment (T()) developed is:
(T ( )) = c T ( )
In which:
(7.1)
c
= expansion coefficient of concrete
T() = temperature increment at time t =
Besides the temperature induced strains, deformations due to swelling and/or hardening
shrinkage might also occur. This issue is discussed in more detail in section 7.6. The hardening
shrinkage, which is almost equal to the autogenous shrinkage aush (see section 7.6.4) plays an
important role for low water cement ratio concretes.
For the total deformation increment of concrete in the hardening stage at time t = it holds:
T + aush = c T ( ) + aush
(7.2)
(t ,=
) ( c T ( ) + aush ( )) Ec ( ) r ( )
(7.3)
where:
T(t)
T ( )
(T ( )) = c T ( )
To
aush
shrinkage
Ec ( )
(t , ) = [ (T ( )) + aush ( ) ] (t , ) Ec ( ) r ( )
()
(t,)
t <
compression
d. stress increment ( )
stress =
( )
time
tension
Fig. 7.1
The mechanism that leads to stress development and cracking in hardening concrete.
Due to the influence of relaxation, stresses are reduced in accordance with (Fig. 7.1d):
(t , ) = [ (T ( )) + aush ( )] (t , ) Ec ( ) r ( )
where:
(7.4)
Stress reduction due to relaxation is an important issue in hardening concrete. Relaxation factors
traditionally applied for the description of the stress development in hardened concrete are not
applicable. The relaxation behaviour of hardening concrete, therefore, requires due attention (see
section 7.8.2)
Summation of the stress increments (t,) results in the overall (tensile) stress ct(t) according
to:
i
ct (t ) = (t , i )
(7.5)
Cracking occurs whenever the tensile stresses ct(t) exceeds the actual tensile strength (Fig. 7.1e).
Expressed as a formula:
P{Fcr}(t)
ct(t)
fct(t)
(7.6)
As mentioned before, the description of the hydration process is the basis for the description of
the behaviour of hardening concrete. The quantification of the hydration process is discussed in
more detail in the following sections.
layer of hydration products around the cement particles becomes thicker and particles are in
contact with each other. After this stage, the process becomes diffusion controlled.
A microstructure develops, existing of unhydrated cement, hydration products (gel) and a
(capillary) pore system, partly filled with water. The gel consists of a solid material and water
filled gel pores. This gel water is physically bound to the surface of the gel particles. After a
certain period, the acceleration stage changes towards the ceasing stage.
1.0
Qmax
degree of hydration
0.5
heat-development Q ( )
degree of hydration
h ()
Q*max
h () =
Q ( )
Qmax
acceleration period
hydratationcurveh (t)
degree of reaction
de-celeration period
dormant stage
r () =
0
Fig. 7.2
h ( ):
r ( )
Q ( )
Q*max
time (t)
h =
originally available amount of cement Vce (0)
(7.7)
When it is assumed that the amount of liberated heat is proportional to the amount of cement that
has reacted, the degree of hydration can also be presented with the ratio:
h ( ) =
where: Q()
Qmax
Q( )
Qmax
(7.8)
V
Vcw
Vg
h =
Vce (0)
= air volume
Vcw
Vg
= gel volume
Vce
Vce
Fig. 7.3
Vl
The amount of heat developed at complete hydration depends on the type of cement. For Portland
cement in particular, the value of Qmax is between 350 and 500 J/g cement. A relatively high
C3A 1) and/or C3S2) content results in a high value of Qmax. A high amount of C2S3) provides a
lower and slower heat development.
Depending on the type of cement, a water-cement ratio (wcr) of about 0,4 is theoretically required
to obtain complete hydration of all the cement present. In practice however, a wcr of 0,4 does not
result in complete hydration. Fig. 7.4 provides an impression of the maximum degree of
hydration h,max expected in practice as a function of the wcr. For finer cements, higher values for
h,max can be achieved and for more coarse cements lower values for h,max will be reached than
the results presented in the figure suggest.
1.0
degree of hydration
0.8
0.6
0.4
0.2
0
Fig. 7.4
1)
0.2
0.4
0.6
0.8
1.0
wcr
r ( ) =
Q( )
Q* max
(7.9)
For Q*max, the amount of heat liberated after 28 days of hardening can be used or, instead, the
value determined in an adiabatic hydration test after e.g. 7 days of hardening. After this period,
the hydration process almost ceases. As a result of this, the sum of the heat produced hardly
changes.
For a certain mixture, i.e. a mixture with a certain wcr, an unambiguous linear relationship
between the degree of reaction and the degree of hydration exists:
r ( ) =
Qmax
h ( )
Q* max
(7.10)
Based on Fig. 7.4, it is seen that, for most commonly used concrete mixtures in practice with
wcrs ranging between 0,40 and 0,55, a degree of reaction of 100% corresponds with a degree of
hydration between 0,7 and 0,8. For low water-cement ratio mixtures, like high strength concrete
(wcr 0,3), the degree of hydration is limited to about 0,5 0,6. The high strength of these
concretes is therefore not a result of the high degree of hydration, but is the result of a high
packing density of the particles.
Ta (t ) =
where:
T(t)
C
c
cc
h (t ) C Qmax
c cc
(7.11)
[C]
[kg/m3]
[kg/m3]
[kJ/kg.C]
To illustrate this, Table 7.1 provides an example of an adiabatic temperature rise in a normal
strength concrete mixture.
temperature [C]
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Tadi
T0
Fig. 7.5
degree of hydration
h (adi)
time
The heat capacity c cc [kJ/m3 C] of hardening concrete is, strictly speaking, not a constant value,
but is a function of the degree of hydration. If, to avoid complexities, the heat capacity were a
constant (which, for most practical applications, is acceptable) a linear relation exists between the
adiabatic temperature rise and the degree of hydration. This implies that both have the same
course in time (Fig. 7.5). For this particular case, the adiabatic temperature rise is often used to
determine the degree of reaction, using a so-called adiabatic calorimeter.
Table 7.1 Example of adiabatic temperature rise calculation
Input parameters:
Heat capacity:
Max. degree of hydration:
Max heat production:
Cement content:
Question:
The shape of the adiabatic curve also depends on the following set of factors:
Chemical composition of the cement
Particle size distribution of the cement (Blaine)
Water-cement ratio
Cement content
Temperature
Availability of agents (accelerators, retarders)
In Fig. 7.6, for three different concrete mixtures, the adiabatic heat development is presented as a
function of time. It concerns a normal strength concrete (C = 350 kg/m3 Portland cement), a high
strength concrete (C = 475 kg/m3) and a Granulate concrete (C = 150 kg/m3 blast furnace slag
cement CEM III/B). The latter two adiabatic curves provide a clear impression of the range of
adiabatic temperature rises to be expected.
Fig. 7.7 provides three adiabatic curves for Portland fly ash cement for three different casting
temperatures. What is striking is that a lower casting temperature initially shows a lower heat
development, but in the longer run shows an increase of the heat production. The cause of this
phenomenon is the influence of the hardening temperature on the structure of the hydration
products formed [2].
80
temperature [C]
b
concrete characteristics
60
a. C25/C35
b. C85
c. granulate concrete
40
c
20
0
Fig. 7.6
24
48
72
96
120
144
168
time [hours]
50
40
TA 53
30
20
T0=30
20 10
10
0
0
Fig. 7.7
1 2345
12
24
36 48
72
96
120
168
time [hours]
h (t)
0.5
Tsemi adi
degree of hydratation
h (t)
1.0
adiabatic curve
time
Fig. 7.8
The course of a semi-adiabatic hydration process, also called the process curve, is determined by
the temperature Tp(t) at which the reaction process takes place. Oppositely, the temperature on its
turn depends, among other things, on the amount of heat produced per unit of time. Thus, the
temperature development and the heat development are coupled quantities.
10
The mutual dependency between the rate of hydration and the concrete temperature implies that
the calculations of both the temperature development and the hardening process are to be carried
out using a step-wise calculation procedure in which the effect of the actual temperature and the
rate of reaction are modelled explicitly.
1
T
2T
q ( x, t )
= ac 2 +
c . cc c
t
x
(7.12)
ac =
where:
ac
c
cc
c
T(x,t)
qc(x,t)
x
t
c
c . cc
(7.13)
[m2/h]
[W/m.K]
[kJ/kg.K]
[kg/m3]
[K]
[kJ/m3.h]
[m]
[h]
With qc(x,t) = Qc(x,t) / t and the transition towards a finite difference approach, the differential
equation is expressed as:
Qc ( x, t )
T
2T
1
+
= ac
2
c . cc
t
t
(x)
(7.14)
where Qc(x,t) [kJ/m3] is the heat development within a time step t. The temperature at time
t+t is found by adding the temperature rise caused by the heat liberation Qc(x,t) to the
temperature T(t). This temperature rise is calculated by dividing the amount of heat Qc(x,t)
liberated within a time step t by the heat capacity c.cc.
For this particular case, the heat increment Qp(x,t), in which the index p implies the process,
can be determined according to (see Fig. 7.9):
(7.15)
11
f ( h; j , Tp; j ) = e
R
Ta; j Tp; j
(7.16)
where:
EA(Tp)
R
Qa;j+1
Qp;j+1
Ta;j
Tp;j
The quantity f(h;j,Tp;j) relates the rate at which the hydration process takes place to the rate at
which the process would take place if the conditions were adiabatical (see Fig. 7.9).
Qmax
Q (t)
temperature (C)
t j+1
Qa;j+1
Ta;j
Tp;j
Qp;j+1
Qa;j
Qp;j
T0
tj
tj+1
t j+1
Fig. 7.9
time
For T 20 C:
For T < 20 C:
EA = 33.5 kJ/mol
EA = 33.5 + 1.47(20-T) kJ/mol
12
Krel
b
EA =50 kJ/mol
EA =35 kJ/mol
EA =30 kJ/mol
a. Danish cement
b. Swedish cements
0.5
0
0
273
Fig. 7.10
EA =45 kJ/mol
10
20
293
30
40
50
60
70 C
313
K
333
temperature
Temperature function Krel = f(h,T) for different values of the apparent activation
energy EA.
A quantitatively comparable result as achieved with formula (7.16) is obtained with the relation:
Ta Tp
f ( h , Tp ) = g
10
(7.17)
where g is a temperature sensitivity factor. This expression is comparable to the one used for
the activation energy EA. The g-value depends on the temperature, the degree of hydration and
the chemical composition of the cement. With g-values between 1,6 and 2,2, in general good
results are obtained.
To solve the differential equation of Fourier with a heat source, solution procedures described in
chapter 2 are used. The temperature rise due to heat liberation in a certain time step and at a
certain location is accounted for in each time step. The result of a temperature calculation in
hardening concrete is given in table 7.2.
13
h(x,t+1)
T(x,t+1)
temperature T(x,t)
degree of hydratation
h(x,t)
The degree of hydration is the most important parameter for the description of the material
properties. These concern the thermal, physical and mechanical properties.
The accuracy of the temperature calculation depends, at given thermal boundary conditions,
on the accuracy when describing the temperature sensitivity of the reaction process and the
development of the thermal properties.
(7.18)
where m is the heat-conduction coefficient of, e.g., the formwork material and the surroundings,
i.e. air.
14
70
temperature ( C)
(W/m2 K)
5
15
25
middle
60
1.0
0.8
50
0.6
40
0.4
0.2
30
top side
7 days
20
0
24
48
72
96
Fig. 7.11
0
20
30
3 days
40
1 day
50
60
70
temperature ( C)
The temperature development is calculated for transfer coefficients of about 5, 15 and 25 W/m2K,
associated with wind speeds of 0, 2 and 5 m/s. Fig. 7.11a gives the temperature development for
different transfer coefficients in time, for both the centre and the top side of the concrete plate.
Fig. 7.11b provides the temperature development over the thickness of the concrete plate at three
different ages, i.e. 1, 3 and 7 days. In order to show the effect of the thermal boundary conditions
most explicitly, a high cement content mixture is investigated.
It is quite clear that the presence of a surface wind results in a significant increase of the
temperature differences in a concrete cross-section. As a result of the temperature decrease at the
surface, the hardening process evolves more slowly. The elastic modulus also develops more
slowly. This is a positive by-effect since it results in a more moderate development of the thermal
stresses in this surface zone. Thanks to the moderate development of the stresses, stress relaxation
is more effective. Quantification of the different influences on the stress development requires the
use of advanced numerical programs.
15
80
surface temperature ( C)
with solar radiation, no formwork
with solar radiation, removal of
formwork after 48 hours
60
40
20
no solar radiation, no formwork
no solar radiation, removal of
formwork after 48 hours
0
0
Fig. 7.12
20
40
60
80
100
120
time (hours)
ac =
c
c . cc
It represents the rate at which the heat transport takes place. During hardening, the structure of the
concrete will change, and, as a result, also the heat conduction coefficient c and the specific heat
cc .
In the literature, contradictory data is given on the change of the heat conduction coefficient c
with elapse of the hydration process. It seems that with an ongoing degree of hydration, a slight
increase of about 10 % looks reasonable.
For the specific heat cc at increasing degree of hydration, a slight decrease is observed.
16
A slight increase of the heat conduction coefficient and a reduction of the specific heat results in
an almost constant value of the temperature leveling coefficient ac. For practical applications it is,
therefore, often allowed to adopt a constant value for the thermal properties of concrete. For
fundamental research, however, the changes in the heat diffusion coefficient during progress of
the hydration process should be accounted for.
Thermal changes
Plastic shrinkage
Chemical shrinkage
Autogenous shrinkage
Expansion of reaction products
Expansion due to moisture adsorption (submerged hardening)
Restraining the expansion caused by the expanding reaction products in the early age of
hardening and the moisture adsorption, results in relatively low compressive stresses. Since the
stiffness of the concrete is still very small at that time, the compressive stresses will also be quite
low. In this report, these stresses will not be considered. Special types of cement have been
developed of which swelling is a main characteristic. An example is the American type K cement.
Swell cements, as well as the phenomenon of swelling, as holds for normal cements, will not be
considered.
17
r.v.
(%
) 10
70 80 90 0
n
co
e te
cr
m
te
pe
t
ra
( C
5 0 60
e
ur
) 40
35
40
30
30
25
20
10
40
35
40
km
/h
)
25
30
20
sp
ee
d(
15
wi
nd
5 10
20
30
air temperature ( C)
10
5
1
0
Fig. 7.13
stress N/mm2
stress N/mm2
fct1
fct
fct2
capillary
capillary
crack development
no cracking
time
Fig. 7.14
time
18
19
Table 7.3
Complete hydration of cement theoretically requires a water cement ratio of 0,4. At hydration of
X grams of cement, 0,15 X grams of water are physically bound and 0,25 X grams are bound
chemically. 0,25 X gram water has a volume of 0,25 X cm3. The volume reduction of the
hydration products with respect to the volume of the individual components is:
V = 0,25 0,25 X cm3 = 0,0625 X cm3
Expressed in volumetric percentages of completely hydrated cement, in which no free water is
available anymore, for the porosity P op the hardened cement it holds:
P
=
V
100%
=
X wcf X
+
ce
w
0, 0625 X
100%
= 8, 6%
X 0, 4 X
+
3,1 1, 0
where ce and w are the specific mass of cement and water, respectively.
Autogenous shrinkage manifests itself mainly for water-cement ratios below 0,4. For a cement
paste with a wcr of 0,4, Grube [13] measured an autogenous shrinkage of about 0,710-3 after 28
days of hardening. For normal strength concrete, autogenous shrinkage is about 1/6 of the
shrinkage of cement paste. In this case, aush = 0,1210-3. This is about the cracking strain of
concrete!
Large autogenous shrinkage occurs in high strength concrete. In Fig. 7.15, values of the
autogenous shrinkage of cement paste and concrete are given for different values of the wcr. In
Fig. 7.16, the autogenous shrinkage is a function of the degree of hydration. There is a strong
correlation between the autogenous shrinkage and the degree of hydration at low wcrs.
20
0
wcr-SF-SP1
0.40- 0- 0
1000
0.30- 0- 0
0.23- 0- 0.6
2000
expansion
(x10-6)
autogenous shrinkage (x10-6)
1000
500
days
21 28 42 70 100
wcr- SF- ad
0.20-10-0.75
0.30-10-0.40
0.30- 0- 0.40
0.50- 0- 0.14
0.23-10- 0.6
3000
legend
SF = silica fume
SP = super plasticizer
0.17-10- 2.0
4000
Fig. 7.15
1 2 4 7
21 42 100
days
Autogenous shrinkage of cement paste (left) and concrete (right) for different wcrs
[24]. (Note: be aware of the differences in scale).
-0.25
-0.20
-0.15
-0.10
-0.05
0
0
0.1
0.2
0.3
0.4
0.5
0.6
degrees of hydration
Fig. 7.16 Autogenous shrinkage of high strength concrete as a function of the degree of hydration
[4]. Note: For any concrete mixture, the autogenous shrinkage might differ
substantially.
Grube [13] remarks that autogenous shrinkage, i.e. shrinkage due to self-drying, also has a
positive effect, in particular when the frost resistance and heat loading are concerned. In those
cases, the self-drying introduces pore spaces that either act as an additional expansion capacity in
case of ice formation or store the expanding water vapor.
21
M = (T (t ) + Tr )dt [C.h]
(7.19)
where T(t) is the concrete temperature at time t and Tr is the reference temperature. For the
reference temperature, Saul adopted a value Tr= -10C. For the maturity as considered, there is a
linear relationship between the strength and the logarithm of the maturity (Fig. 7.17b).
temperature ( C)
fc cm
t
T (t)
T
0
-10
time
Fig. 7.17
n (maturity) (C.h)
b. Strength as a function of the logarithm of maturity
22
temperature ( C)
30
20
10
0
-10
0
Fig. 7.18
n
n
n
n
nnnn-
1.5
1
0.5
0
0.5
1
1.5
2
20
40
time of hardening (hours)
Graph with n-values used to calculate the weighted maturity Mw according to the
CEMIJ-method [11].
Table 7.4 C-values for Dutch cements (method CEMIJ) (e.g. [11]).
Cement type
Notation
Strength
Characteristics
C-value
Blast furnace slag
cement
Blast furnace slag
cement Blast furnace
slag cement Blast
furnace slag cement
Portland cement
Portland cement
Portland cement
CEM III/B
CEM III/B
CEM III/B
CEM III/B
42,5
42,5
42,5
42,5
CEM I
CEM I
CEM I
CEM II- B-V
LH HS
LH HS plus
LH HS
Old cement
notation
1,65
1,65
1,50
1,45
A
B
A
B
32,5
42,5
52,5
1,30
1,30
1,30
A
B
C
32,5
1,30
the Cement plant IJmuiden (CEMIJ) and is known as the method De Vree (see [11]).
According to the CEMIJ-method, the weighted maturity Mw is directly determined using an
expression that replaces the calculation time intensive integral from equation (7.19) (see Fig.
7.18):
23
M w = On .C n [C.h]
(7.20)
where On is an integral surface increment [C.h] to which a value n is linked. The n is the power
of the C-value in the calculation of the weighted maturity. In table 7.4, different indicative Cvalues are given. By using these so-called weight values in the calculation, there is a linear
relationship between the (weighted) maturity and the strength.
Application in practice
The maturity method (CEMIJ-method) is described very accurately in CUR-recommendation 9
[12]. The potential of the method comes from its easy application. After having determined the
calibration curve, the temperature is measured, followed by a calculation of the (weighted)
maturity and, finally, the strength. The strength follows from the calibration curve. For a more
extensive description of the maturity method, reference is made to the literature [11,12,14].
Remarks:
1. The term maturity is sometimes used to indicate the degree of hydration [15]. By doing so, things become
obscured. The maturity is a parameter with dimension [C.h], whereas the degree of hydration is a dimensionless
quantity that ranges from 0 to 1,0, or 0 to 100% respectively.
2. Instead of using the measured temperature to determine the maturity, the calculated temperature can also be used.
In this way, one can simulate all kinds of boundary conditions in the design stage of a project. The results might
be a basis for the choice of a concrete mixture, casting procedure or time of removal of the formwork.
f cm (t ) = cc f cm
(7.21)
where:
cc (t ) = e
where: fcm(t)
fcm
cc(t)
t
t1
s
28
s 1
t / t1
0,5
(7.22)
These formulae are valid within the temperature range -20C to + 40C. For temperatures within
the range from 0C to 80C, a maturity method is given. This method proposes an equivalent time
(in Dutch: equivalente tijd) according to:
24
t (T ) = ti e
4000
13, 65
273+T ( ti ) / T0
(7.23)
i =1
where:
t(T)
T(ti)
ti
T0
[d]
[C]
[d]
[C]
100
wcr = 0.3
wcr = 0.4
80
wcr = 0.5
60
wcr = 0.6
40
20
0
0
Fig. 7.19
0.2
0.4
0.6
0.8
1.0
degree of hydration
25
7.7.3.1b
Practical applications The UCON-system
Application of the degree of hydration concept to determine the strength development, presupposes that the degree of hydration as well as the relationship between the degree of hydration
and the compressive strength are known. The degree of hydration is relatively easily determined
with the help of special purpose software. The input required for a calculation consists of:
1. The adiabatic hydration curve of the concrete mixture;
2. The concrete temperature during hardening.
That the degree of hydration can be determined using these two input data can easily be seen if
the algorithm, which has been discussed earlier in section 7.4.4 for the determination of the
temperature development, is considered. The calculation algorithm is explained with the help of
Fig. 7.9. In Fig. 7.20, figure 7.9 is shown again, almost without any changes. When explaining
Fig. 7.9, it was stated that the heat increment Qp;j+1 within time step tj+1, can be calculated if
the calculated temperature from the previous time step Tp;j is known. Instead of this calculated
temperature, the measured concrete temperature can be used too. Based on this measured
temperature and the adiabatic hydration curve, exactly the same calculation can be conduced as
described in section 7.4.4. The result of the calculation is the degree of hydration as it develops at
the prevailing temperature conditions within the concrete!
Qmax
Q (t)
1.
temperature (C)
t j+1
2.
Qa;j+1
Ta;j
Tp;j
3.
Qp;j+1
Qa;j
Qp;j
2
T0
tj
tj+1
t j+1
Fig. 7.20
time
Calculation procedure for the determination of the degree of hydration h(t) from the
adiabatic hydration curve and the measured concrete temperature Tp(t).
The software for calculating the degree of hydration from the adiabatic hydration curve and the
measured temperature is part of the UCON-system (Universal Concrete Curing CONtrol system)
developed at the TU Delft [25,29]. Fig. 7.21 presents the UCON-system schematically. The
adiabatic hydration curve is determined experimentally (or numerically [28]) and is imported in
the computer. At pre-assigned locations within the structure, temperature measurements are
performed. Their results are stored in the computer. Based on the imported data, the degree of
hydration is determined for all these locations. For a known and a priori data base stored
relationship between the degree of hydration and the strength, the strength in these particular
locations is determined.
The relationship between the compressive strength and the degree of hydration is determined
quite easily, using results from compressive strength tests on concrete cubes hardened in
26
compliance with a known predefined curing temperature. With assistance of the UCON-system,
the degree of hydration that corresponds with the temperature profile is determined. By
measuring the compressive strength at different times, i.e. different degrees of hydration, the
relation between the degree of hydration and the compressive strength is determined for a specific
concrete.
If the relation between the degree of hydration and the strength is not determined in advance, use
is made of a graph as shown in Fig. 7.19 or by consulting other relations for similar concrete
mixtures from the literature.
It is noted that the relation between the compressive strength and the degree of hydration for high
strength concrete as proposed by Fagerlund, shows and underestimation of the compressive
strength [17]. The reason for this is most likely the effect of silica fume in the concrete. This
aspect was at that time not considered by Fagerlund.
controller
thermocouple
24 72 168 672
concrete
strength
adiabatic bath
computer
adiabatic curve
digital
thermometer
information
temperature
concrete
structure
thermocouples
Fig. 7.21
27
90
concrete temperature ( C)
80
0.60
adiabatic curve
0.50
70
60
0.40
50
0.30
40
measured temperature
in center of cube 150x150x150 mm
30
20
Fig. 7.22
0.20
0.10
10
0
0
degree of hydration ( h)
10
20
30
40
50
60
time (hours)
0
0
10
20
30
40
50
60
time (hours)
Fig. 7.22 shows the result of a degree of hydration calculation with the UCON-system. It
concerns a high strength concrete project. The concrete has hardened under isothermal conditions
(process temperature Tp(t)). The adiabatic hydration curve is determined experimentally.
++
+
+
0.4
0.5
0.6
Fig. 7.23
+
0
20
40
60
80
degree of hydration (%)
Tensile strength as a function of the degree of hydration for different wcrs. Blast
furnace slag cement CEM II. Cement content: 350 kg/m3 [19].
28
Ec (t ) = E (t ) Ec
(7.24)
E (t ) = [ cc (t )] 0,5
(7.25)
where:
where:
Ec(t)
Ec
E(t)
cc(t)
Quite often, the elastic modulus is determined using existing relations between the elastic
modulus and the compressive strength, e.g. VBC (NEN 6720), CEB-FIP Model Code 1990 and
EN 1992-1-1. In the Model Code, it holds:
Ec (t ) = 9500 [f ck (t ) + 8]
1/ 3
(7.26)
where fck(t) is the characteristic concrete compressive strength (prism strength) at time t.
For a more advanced discussion on this issue, use can be made of the theory in which concrete is
a multi-phase material existing of aggregates and cement paste. In this case, the cement paste is a
two-phase system containing un-hydrated cement and hydration products. The resulting elastic
modulus is expressed as a function of the degree of hydration [19].
50000
40000
0.4
+ +
0.5
30000
0.6
20000
10000
0
0
Fig. 7.24
20
40
60
80
degree of hydration (%)
Course of the elastic modulus as a function of the degree of hydration for the
different wcrs. Cement content 350 kg/m3, Blast furnace slag cement CEM II [19].
29
For the relation between the elastic modulus of concrete Ec and the degree of hydration h, Gutsch
proposes [18]:
Ec ( h ) = E0
where:
E0
0
h 0
1 0
(7.27)
Fig. 7.24 provides an example of the calculated course of the elastic modulus as a function of the
degree of hydration for different wcrs.
(ti , j ) = r (ti , j ) c ( h ) T ( j ) Ec ( j )
with:
(ti,j,)
j
r
(ti,j,)
c(h)
T(j)
Ec(j)
(7.28)
30
compression
tension
age
(t) = E
relaxation
Fig. 7.25
The resulting stress at time t is found by adding the stress increments as discussed earlier in
section 7.2.
7.8.2 Relaxation
Relaxation has a considerable influence on the stress development. Exact information about this
issue is however very scarce. Some programs take into account the effect of relaxation by
reducing the elastic modulus by 50% during a predefined period. Due to the lack of information,
relaxation is sometimes completely neglected. A relaxation formula that gives acceptably
accurate results for the stresses in hardening concrete holds [26]:
(t , ) = e
with:
n
h
wcr
( t )
(t )
h 1 + 1,34 ( wcr )1,65 d ( t )n h
h ( )
h ( )
(7.29)
With the quotient c/h, the effect of ongoing hydration during the period (t-), is taken into
account in the relaxation formula. If hydration stops (which is when c/h = 1), formula (7.29)
reduces to:
(t , ) = e
(7.30)
31
or:
(t , ) = e a
( t )n
(7.31)
or:
(t , ) = e (t , )
where:
(t,)
a
(7.32)
Fig. 7.26 gives an impression of the relaxation factor (t,) as a function of time. In this figure,
also the course of the degree of hydration is presented. It can be observed clearly that stresses that
develop during the early acceleration stage of the hydration process almost completely vanish.
A disadvantage of the superposition method is the fact that is requires substantial memory
allocation and calculation time. For a one-dimensional problem, the current pcs work out well,
but for the calculation of three-dimensional stress fields, different relaxation formulas
1.0
relaxation factor
(t,)
relaxation curves calculated
with formula (7.29)
0.8
0.6
48 h
0.4
34 h
degree of hydrationh (t)
0.2
24 h
19 h
0
0
Fig. 7.26
40
80
120
160
200
time (hours)
Relaxation curves (t,) for young concrete, calculated with formula (7.29).
should be adopted (rate type). By doing so, substantial computation time reductions can be
reached at the cost of the availability of the stress history that the hardening concrete
experiences. Many multi-dimensional practical problems can be reduced to a one-dimensional
problem and can then be solved using the superposition method.
32
surface centre
Fig. 7.27
stress (N/mm2 )
8
7
6
5
4
3
2
1
0
-1
-2
-3
-4
-5
0
20
40
r = 1.0
r = 0.5
r = 0.25
r=0
increasing degree
of restrained
60
80
100
Both longitudinal and bending deformations (curvature) of the wall are not restrained
(rl = 0,0; r = 0,0)
Longitudinal deformations are not restrained (rl = 0,0), but the bending deformations
of the structure are restrained completely (r = 1,0)
33
6.0
5.0
r1 = 0.0
r = 0.0
+
4.0
5.00m
r1 = 0.0
r = 1.0
case A: rl = 0 ; r = 0
3.0
2.0
0.70m
case B: rl = 0 ; r = 1,0
1.0
0.80m
0
-2
-1
0
1
stress (MPa)
4.10m
Calculated stresses (after 5 days; cooling stage) over the height of wall-slab structure,
for different values of the degree of restraint for the length (rl) and bending
(r)deformations.
The result of the stress calculations is in Fig. 7.28b. There, the stresses are over the height of the
wall after 5 days of hardening. It becomes clear that for case A, where the wall is free to deform
(rl = 0,0), i.e. not restrained in any way, the stresses just above the connection with the floor slab
are larger than in case B, where the wall is fully restrained (r = 1,0). At the top of the wall, case
A shows compression whereas for case B, the wall as a whole is in tension.
What happens here is that the course of the deformation of an individual fiber (micro-level) is
completely different from the course expected when considering the assumed degree of restraint.
The degree of restraint is a parameter that defines the deformational capabilities of a structure as a
whole, not the course of deformation of any individual fiber. The deformation of an individual
fiber is the net result of the imposed (thermal) strain at fiber level plus the strain that comes from
the deformation of the structure as a whole!
Another complication when regarding the issue of the degree of restraint is that the level of
restraint of the deformations changes during the progress of the hydration process. The
deformation of a wall, cast on a rigid foundation slab, is initially restrained completely at the
connection with the slab (the stiffness of the wall is still negligible with respect to the slab: r =
1,0). After a certain time, depending on the stiffness ratio between the wall and the foundation
slab, a part of the imposed deformations results in a deformation of the structure as a whole. Only
this part results in stresses. Finally, the magnitude of the effective imposed strain at fiber level is
determined by the equilibrium conditions and the deformation compatibility conditions in a
cross-section!
34
tensile trajectories
almost free
crack directions
r ~
~1.0
45
Fig. 7.29
45
Deformations of a wall cast on an already hardened floor slab. Indicated are the
degree of restraint, the direction of the tensile trajectories and the direction of the
cracks.
Table 7.5 presents the degree of restraint of walls as a function of the length/height ratio. The
values in the table are indicative values to obtain a first estimation of the degree of restraint of the
deformation at the topside of the wall. At the connection interface with the slab, the degree of
restraint is about 100%, also for short walls (see Fig. 7.29).
In the British Standard 8007-1987: Design of concrete structures for retaining aqueous liquids
[33], some indicative figures are presented for different types of structures. They also give values
for the degree of restraint. These are represented by the figures 7.30a-d (EN 1992-3).
Table 7.5
Degree of restraint as a function of the length/height ratio (L/H) of the wall (after
[21]) (Note: values for a low L/H-ratio are only valid for the degree of restraint at the
top of the wall)
>8
Degree of restraint
0,1
0,6
1,0
35
...
c. Wall fixed at two sides
Fig. 7.30
Degree of restraint r for walls and slabs at different boundary conditions (after BS
8007-1987 [33] & EN 1992-3; no relaxation included). (Note: The effective degree of
restraint also depends on the amount of reinforcement applied!).
36
compression
especially with regard to the standard deviation of the tensile strength!. (see also section 7.8.4.2.).
If the strength and the stress have a normal distribution function, the difference between them, i.e.
z = fct ct, is also normally distributed. The likelihood that at a certain time z is smaller than
zero, which implies that cracking appears, is determined with help of basic statistics (see also the
example in table 7.6).
stress =
()
tension
time
Fig. 7.31
Development of the tensile strength and stress in time (schematically). Strength and
stress are considered to be normally distributed quantities.
Table 7.6
Tensile strength and tensile stress normally distributed. For the difference u(z) = fctm - ct it
holds:
u(z) = fctm - ct = 3,0 2,6 = 0,4 MPa
s(z) = [s2(fct) + s2(ct)]0,5 = 0,396 MPa
Ratio u(z)/s(z) = 0,4 / 0,396 = 1,007
With help of the table for standard normally distribution functions, the probability of cracking is
[30]:
Cracking probability p{Fcr} = 15,8%
Figure 7.32 gives the result of a calculation on the probability of cracking according to the
procedure as described before. The development of the probability of cracking in time is given for
walls that have different thickness. They are cast on an already hardened floor slab. Deformations
are restrained by the floor slab. The wall-slab structure can bend freely (r = 0). The cracking
probability is determined in the most critical cross-section, which in this case is about 0,5 m
above the wall-slab connection. Based on the imposed boundary conditions, the 1,0 m thick wall
almost certainly cracks. The 0,5 m thick wall has a cracking probability of more than 50%. As a
result of the bending of the wall, compressive stresses appear at the top of the wall. The reduction
37
of the probability of cracking after having reached a maximum value comes from the gradual
decrease of the temperature differences in the cross-section.
100
symmetrical axis
centre, d=0.5m
centre, d=1m
80
surface, d=0.5m
surface, d=1m
3m
60
40
surface
centre
20
1m
centre, d=0.25m
3m
Fig. 7.32
0
0
20
40
60
80
100
38
-2
stress (N/mm2 )
cement kg/m 3
wcr 0.5
-1
0
PC-A
HOC-A
1 PC:
0.9 f ctmspl
2
HOC: 0.9 f ctmspl
3
0
Fig. 7.33
24
48
72
96
120
time (hours)
Experimental results
Tests carried out on concrete hardened under semi-adiabatic conditions and experiencing a partly
of full degree of restraint, has shown that cracking occurs at a stress level equal to (on average)
0,85 fctm 2) (see Fig. 7.33). For this tensile stress, the probability of cracking should be 50%. If
both the strength and stress have a standard deviation of 10% of their average values, a cracking
probability of 50% is associated with a stress level equal to 0,85 times the average axial (short
time) tensile strength. This can only be the result if we take into account a reduction factor for
tensile strength of 100 85 = 15%. This reduction of the tensile strength is the result of the fact
that the stresses develop slowly and that, therefore, the long term-effect (reduction) on the tensile
strength must be taken into account.
Cracking probability and safety factors
Given the fact that in many cases hardening concrete cracks at a stress level which is on average
15% below the average axial short term-tensile strength, the following argumentation can be
followed. Assume that for a particular case the tensile strength is 3 N/mm2. In table 7.7, the
probability of cracking is found for a reduction of the tensile strength of 0% and 15%,
respectively. The cracking probability is calculated at values for the average tensile stress ctm of
0,5 fctm ; 0,6 fctm ; 0,7 fctm and 0,85 fctm. Standard deviations of 10% of the average tensile strength
and tensile stress are assumed. For no reduction of the tensile strength, it is observed that at a
stress level ctm = 0,85 fctm , the probability of cracking is 15,8%, whereas the safety factor is
already below the value 1, namely = 0,83. In order to be sure to achieve a safety factor above
1,0, the stress should not exceed 0,7 fctm. For this particular case it holds that = 1,03.
In reality, as has been observed in tests, a reduction of the tensile strength of 15% is to be taken
into account. When assuming a tensile strength 0,85 fctm, at an average tensile stress level ctm =
0,85 fctm, the probability of cracking is 50%. The accompanying safety factor then is = 0,72. At
a tensile stress level of 0,7 fctm, the probability of cracking is 7% and the safety factor against
cracking is still below 1,0, namely 0,88. In this example, the turning point at which the safety
factor reaches a value of more than 1,0 is ctm 0,6 fctm. The probability of cracking then is about
1%. If stresses over 0,5 fctm are not allowed to occur, the safety against cracking is 1,22 at a
cracking probability of 0,02%.
) fctm = average axial tensile strength. It holds that: 0,85 fctm = 0,85 * 0,9 fctmspl = 0,76 fctmspl (see also section 4.3.6.2)
39
Table 7.7
Probability of cracking and corresponding safety factors for different values of the
effective tensile strength.
Quantities
Values
w.r.t tensile
strength
Short term
tensile
strength
Reduced
short term
tensile
strength*)
Stress at cracking
0,5 fctm
0,6 fctm
0,7 fctm
0,85 fctm
Average value
N/mm2
3,0
1,5
1,8
2,1
2,6
Standard deviation
N/mm2
0,3
0,15
0,18
0,21
0,26
Characteristic value
N/mm2
2,51
1,75
2,1
2,44
3,03
Prob. of cracking
[%]
--
0,03%
0,7%
15,8%
---
1,4
1,2
1,03
0,83
Average value
N/mm
2,6
1,5
1,8
2,1
2,6
Standard deviation
N/mm2
0,26*)
0,15
0,18
0,21
0,26
Characteristic value
N/mm2
2,14
1,75
2,1
2,44
3,03
Prob. of cracking
[%]
0,02%
0,6%
7%
50%
---
1,22
1,02
0,88
0,71
*) Tensile strength with a reduction of 15% in relation to the short-term tensile strength fctm = 3,0 N/mm
1.
Note
If the analysis concerns concretes with a very low water cement ratio (for example high strength concretes with a
wcr < 0,35), not only the calculation of the tensile stresses due to temperature effects, but also the effect of
hardening shrinkage should be considered (see section 7.6.4).
2.
For the values of the probability of cracking and the safety factors as calculated, it is assumed that the stresses are
determined at correct thermal and mechanical boundary conditions. Significant deviations of the boundary
conditions, i.e. unexpected variations of the surrounding temperature, high wind speeds etc., are not included. To
gain insight into the effects of these types of variations on the probability of cracking, additional sensitivity
analyses must be performed.
When summarising the results, it can be stated that tensile stresses during hardening limited to about 50% to 60% of
fctm, result in a safety against cracking higher than 1,0. The probability of cracking then is negligible. If tensile stresses
up to 70% of fctm are accepted, cracking is very likely to occur.
40
For the eigen stresses (see chapter 2) in particular, the 20C criterion is sufficient as long as
the wall thickness is not over 1,0 m [20]. For higher wall thicknesses, problems might arise.
Whether cracking occurs depends on the age at which the temperature differences develop
and why they develop. A cold wind passing alongside a relatively warm wall from which the
formwork is removed, makes it relatively sensitive to cracking.
For a wall cast on a hardened foundation, the order of magnitude of the degree of restraint is
of dominant importance with regard to cracking. The magnitude of this degree of restraint
depends on the actual location in the structure (see also [20,34]).
The conclusion to be drawn from these studies is that temperatures and temperature differences
in a number, but certainly not all, cases can be considered as providing a reliable cracking
criterion.
41
7.9 References
1. Mills, R.H. 1966. Factors influencing cessation of hydration in water-cured cement pastes,
ACI-SP, pp. 406-424.
2. Breugel, K. van, 1991. Simulation of hydration and formation of structure in hardening
cement-based materials. PhD-thesis TU-Delft, 295 p.
3. Laube, M. 1991. Werkstoffmodell zur Berechnung von Temperaturspannungen in massigen
Betonbauteilen im jungem Alter. PhD-thesis, Braunschweig.
4. Koenders, E.A.B., e.a. 1995. HSB-kantoor te Brade. Stevin report, TU-Delft, 109 p.
5. Saul, A.G.A., 1951, Principles underlying the steam curing of concrete at atmospheric
pressure. Magazine of Concrete Research, No. 6.
6. Bergstrm, S.G., 1953. Curing temperature, age and strength of concrete. Mag. of Concr.
Res., pp. 61-66.
7. Plowman, J.M., 1956. Maturity and strength of concrete. Magazine of Concrete Research,
pp. 13-22.
8. Papadakis, M., Bresson, J., Contribution letude de facteur de maturit des liant
hydrauliques ; Application l industrie du bton manufactur. CERIB-publications
Technique, No. 8.
9. Bresson, J., 1980. Prevision des resistances: facteur de maturit et temps equivalent.
Annales, No. 387, pp. 106-110.
10. Carino, N.J. e.a., 1983. Temperature effects on strength-maturity relations of mortar. ACIJournal, No. 80-17, pp. 177-182.
11. Gewogen rijpheid, 1984. Betoniek 6/20.
12. CUR-aanbeveling 9: Bepaling van de sterkteontwikkeling van jong beton op basis van de
gewogen rijpheid. Gouda, 1987, 7 p.
13. Grube, H. 1990 Ursachen des Schwindens van Beton und Auswirkungen auf Betonbauteile,
Habilitationsschrift, Darmstadt, 81 pp.
14. Dekker, I., 1995. Research into validity-sphere of maturity-method de Vree. Ijmuiden,
50p.
15. Hansen, T.C., 1970. Physical composition of hardened Portland cement paste. ACI-Journal,
pp. 404-407.
16. Fagerlund, G., 1987. Relations between the strength and degree of hydration and porosity of
cement paste, cement mortar, and concrete. Research report.
17. Toepasbaarheid van beton met hoge sterkte in de 2e Stichtse Brug. Onderzoek in opdracht
van RWS. TU-Delft, 1996.
18. Gutsch, A., Rostsy, F.S., 1994. Young concrete under high tensile stresses Creep,
Relaxation and Cracking. Proc. RILEM Symp. Thermal Cracking in Concrete at Early
Ages, Munich, pp. 111-118.
19. Lokhorst, S.J. et al. 1996. Ontwikkeling van de mechanische eigenschappen: sterkte en
stijfheid. In: Het grijze gebied van het jonge beton. Artikel series in CEMENT 1995/1996.
20. Eberhardt, M., 1993. Rissgefahr in jungem beton, Delft/Darmstadt.
42
43