an cold thermal
TRANSCRIPT
-
7/27/2019 An Cold Thermal
1/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 1
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS:
THEORY AND APPLICATIONS
N. Vitharana1
and R. Wark2
ABSTRACT
Large concrete placements such as those encountered in dam construction are subjected to severe stress
conditions at early age due to the heat generated in the process of hydration of cement. These thermal stresses
can be higher than those experienced by the structures during its service life. Being young concrete, its tensile
strength is much lower than that of hardened concrete. Consequently if measures are not taken, thermal stresses
could lead to the cracking of concrete.
In the traditional approaches, the criterion of limiting the maximum temperature rise (in some cases both
average and differential) is specified as the sole criterion to avoid thermal cracking. However, practical
experiences has shown this approach to be superficial, being either conservative or unconservative dependingon the conditions. This is due to the fact that thermal cracking occurs when thermal stresses exceed the current
tensile strength of the concrete and accordingly temperature limits have no direct relevance. Also, traditional
deemed-to-satisfy criterion of limited maximum temperature rise, based on experience(s) at past projects, would
not be valid for todays conditions where cement types and construction techniques are very different.
Consequently, some modern design and construction practices, such as Japanese standards, permit or require
designers and contractors to develop their own procedures/criteria with respect to thermal crack control in
concrete structures.
A method involving the calculation of a thermal cracking index (ratio of thermal stress/tensile strength) would
be a better and more rational approach. However, this method needs to consider structural, hydration, material,
thermal and exposure parameters. As early-age concrete is in a semi-plastic state with the involvement of many
interactive parameters such as creep, temperature, ambient conditions, strength gain etc., the evaluation of the
thermal cracking index entails complex procedures.
This paper presents the details of a numerical procedure (THERMAL) developed to predict the time-history of
thermal cracking index. Examples are also presented to show where this procedure has been successfully
applied.
Keywords: Thermal stresses, heat-of-hydration, concrete placements, creep, young concrete, strength gain.
1Principal Engineer, GHD Pty Ltd, BSc(Eng)Hons, PhD(Struct), MBA, PG-Dip(Geotech), MASCE
2Technical Director (Dams), GHD Pty Ltd, BEng(Hons),BAppSc(Maths), MEngSc, FIE(Aust)
1 INTRODUCTION
Hydration-induced cracking in concretestructures, particularly in dams and water-
retaining structures, could result in multifariouseffects: accelerated corrosion of reinforcing andprestressing steel, unsightly appearance, loss of
water-tightness, accelerated deterioration due tofreezing-thawing and alkali-aggregate reaction,development of hydrostatic pressures insidecracks in dams, impairment of structuralintegrity, stability and load redistributions. Thenegligence or cursory treatment of early-agestresses could incur heavy costs requiringextensive repair work or even the total early
replacement of structures such as that occurred
with Kinzua dam stilling basin [HOLLAND,1991].
Engineers are well conversant with the analysisand design of structures against applied loadssuch as dead and live loads because appropriateanalysis methods and design criteria are welldocumented. Deformation-induced loadings
such as temperature and shrinkage are treatedwith rule-of-thumb or deemed-to-satisfy
approaches, if considered at all. The adoptionof a minimum reinforcing steel ratio (AS 3600)to avoid early-age thermal cracking in massivestructures is questionable as bond transfer inyoung concrete is very limited [VITHARANA,
1995c].
-
7/27/2019 An Cold Thermal
2/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 2
A popular method in tender specifications forthe construction of dams and large structures is
to place limits on the temperature rise anddifferential. This is achieved by reducing theconcrete placing temperature, internal andexternal cooling, multi-lift concrete placementetc. In recent decades, the introduction ofdifferent cement types and constructionmethods requires the validity of thesetraditional approaches to be assessed.Moreover, the cracking of young concreteoccurs when the thermal stresses exceed theavailable tensile strength at a given timefollowing the casting of concrete. Therefore,the adoption of the traditional method of limited
temperature rise could lead to either superfluous
or inadequate measures to guard against thermalcracking of concrete structures.
The tendency to cracking at an early age inconcrete structures is determined by the ratio of
thermal stress/tensile strength of the hardeningconcrete. The generation of thermal stresses
depends on many factors, mainly: cement typeand content, placing temperature, ambientconditions, concrete material properties,maturity of concrete, structure type,construction sequence and restrained
conditions, creep characteristics etc. Unlike thecase of applied loadings, thermal stresses are
relaxed by the creep effects (non-elasticadditional strains under sustained stresses)which are high in young concrete due to itssemi-plastic state. Also, it is important toconsider construction sequences as this wouldmodify both the thermal and structuralresponses of a structure. Therefore, theevaluation of thermal crack occurrence at early
ages entails a complex procedure. In recenttimes, constitutive models for young concrete
have been developed [THERMAL, 1995]although there is a great need to understandtheir behaviour at fundamental levels.
During 1994-1996, the first Author was on anindustrial fellowship at the HokkaidoDevelopment Bureau, in the most NorthernIsland of Japan. This organisation isresponsible for providing construction advice toall construction activities on the Island.Consequently, it has been conducting research
and development activities in conjunction withmajor contractors particularly for cold weather
conditions as winter temperatures can be as lowas 40
oC and construction periods can be as
short as six months. It has been developing
thermal stress prediction models for RCC damsas early as the 1970s. The temperature and
stress development predicted for a 70m highRCC dam with interruption for the winter isshown in Fig. 1. With such models, it is easierto observe the effect of various parameters anddetermine appropriate measures to be takenwhen construction conditions are varying. TheAuthor was involved in the development ofconstitutive models for young concrete(including large scale tensile creep tests for damconcrete) and a methodology for predicting thetendency to cracking under heat-of-hydration,drying and autogenous shrinkage. Thecomputer program developed [THERMAL,
1995] is suitable for 3-dimensional structures.
(a) Temperature prediction
(b) Thermal stress prediction
Fig. 1 Temperature and stress predictions for a70 m high RCC dam
2 MECHANISM BEHIND
EARLY-AGE THERMAL
CRACKING
Many interactive factors are involved in the
early-age thermal cracking of concretestructures (eg, hydration, thermal, material,
environmental and structural). Therefore, it isdifficult to make firm conclusions on theinfluence of each factor as they depend on a
-
7/27/2019 An Cold Thermal
3/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 3
given situation. The major difference betweenapplied and thermal loadings is that thermal
loading depends on the member stiffness (ie,Youngs modulus of elasticity). Unfortunately,this has not been recognised in the design ofreinforced concrete structures subject toambient thermal loading where the structuralengineers have been misled and considerthermal stresses under ultimate limit stateconditions in which the member stiffness isvery small (being in a plastic stage) resulting innegligible thermal stresses [VITHARANA,1998].
During the hydration phase of concrete, semi-plastic concrete has a very low value forYoungs modulus of elasticity. The
incremental compressive thermal stress, due totemperature change Tt within a time step T,is given by: t = Et Tt where is thecoefficient of expansion for concrete and Et isYoungs modulus of concrete at an age t sincecasting concrete. This stress is further reducedby the concurrent creep which is high for young
concrete. The temperature rise takes placewithin the first few days and the resulting
compressive stresses would be about 1 MPa intypical applications. Once the rate of hydration
retards, the temperature begins to drop (orcooling phase) depending on the rate at whichheat is lost to the surrounding. The temperaturedrop takes place under an increased Et valueand the net stress therefore becomes tensile. Ifthis tensile stress exceeds the available tensilestrength, cracking would occur. Thismechanism is schematically shown in Fig. 2.For comparison, the development of thermalstresses under the same temperature cycle inmatured concrete with a constant E value is alsoshown. As can be seen with constant E, the
induced stress at the end of the temperaturecycle would return to zero.
This highlights that if an accurate assessment ofthe tendency to cracking is required, it isnecessary to consider both the heat generatingcharacteristics as well as the material and
strength properties of young concrete. Inparticular, creep effects should be considered as
it would reduce the magnitude of initialcompressive stresses as well as the subsequenttensile stresses.
Fig. 2 Mechanism behind early-age cracking
(a) Temperature and Youngs modulus
(b) Tensile stress and tensile strength
3 HEAT-OF-HYDRATION
In recent decades, finite element methods have
been developed for heat-transfer analysis and
structural analysis with various levels of
sophistication from simple elastic methods to
non-linear methods [CRICHTON, 1999].
However, little attention has been paid to the
use of appropriate hydration models. The
generation of thermal stresses depends on both
the rate and the total heat generated during the
hydration process. In most thermal stress
modellings, an adiabatic hydration model (in
which heat transfer is not lost to/or gained fromthe surrounding) is used in line with the
traditional approaches. Typical adiabatic
temperature generation curves are given in
[ACI207, 1973] for different placing
temperatures (low to high-heat cements) used
prior to the 1960s and these had been used often
in temperature predictions.
Hydration of cement is a thermally-activated
process and consequently the rate of cement
hydration depends on the reaction temperature
(ie, at which the reaction takes place). Thereaction temperature in turn depends on the
heat-transfer characteristics within the concrete
-
7/27/2019 An Cold Thermal
4/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 4
mass. Therefore, heat-of-hydration and heat-
transfer processes should be coupled in which
the time-history of the reaction temperature is
taken into account and such analysis is known
as coupled or non-linear heat-of-hydrationanalysis [HARADA, 1991]. As will be shown
later, the arbitrary use of adiabatic hydration
models could result in significant errors in
predicted temperatures.
3.1 HYDRATION
CHARACTERISTICS
3.1.1 Adiabatic models
JCI[1986] covers a wide range of cement typesand blend ratios and parameters are given for
developing adiabatic curves for different
cements placed at different temperatures.
These models were calibrated against several
hundred concrete placements throughout Japan.
The adiabatic temperature rise of concrete can
be generally described by:
)1()()(t
t eTT
= (1)
where T(t) = temperature rise (oC) at time t since
casting concrete, T() = total (ultimate)temperature rise directly proportional to the unit
cement content S (kg/m3
of concrete), and =hydration constant representing the rate of
hydration dependent only on the concrete
placing temperature To for a given cement type.
The cumulative heat generated Q(t) (J/m3) up to
time t, for a given cement content S is given by:
)()( tt TcQ = (2)
where c = specific heat of concrete (J/kg/oC),
and = mass density of concrete (kg/m3
). Therate of heat generation, which is to be used in
transient heat-transfer analysis, Qo (J/m3/s) is
given by:
)()()()( t
to
TTceTcQ ==
(3)
Typical values of and T() for differentcement types (eg, ordinary Portland cement,
moderate-heat Portland, Portland fly ash,
Portland blast-furnace slag, and high-strength
concrete) are given in [JCI, 1986] and these are
very useful for design and constructionengineers to make a preliminary assessment of
the heat generating characteristics of a given
concrete mix design. Heat-of-hydration
adiabatic calorimeter tests or large concrete
blocks with insulated surfaces can alternatively
be used when an accurate evaluation of the
parameters is required.
3.1.2 Non-adiabatic models
Adiabatic hydration models give the fastest rate
of heat generation as they are based on the
implicit criterion that the heat already produced
by hydration accelerates this hydration process
in turn. Although this may be valid to the
interior of massive structures such as dams, it
would become invalid near the surfaces of
massive structures or in other members such as
walls, foundations etc.Non-adiabatic hydration models should
consider the time-history of the reaction
(process) temperature. As the reaction
temperature varies within a structural member
due to the concurrent heat-transfer, the
hydration process at a given time varies within
the structural member. In recent times,
sophisticated hydration models have been
developed and tested [HARADA, 1991], but
they are not suitable for routine applications. A
simplified model can be developed from themethod suggested by RASTRUP [1954] based
on heat measurement on cement samples under
constant-temperature (isothermal) conditions
(ie, heat-of solution tests). Supplemented by
laboratory testing and field measurements, it
was shown that [VITHARANA, 1995c] this
model can be easily calibrated and incorporated
in heat-transfer analysis.
With this model, the reaction temperature is
taken into account by relating the actual
reaction time t to an equivalent (or maturity)age te which is determined based on the time-
history of the reaction temperature. The
cumulative heat generated Q(t) (kJ/kg of
cement) up to the reaction time t (days) is
expressed as:
}][exp{)(n
et tmBAQ+=
dtt rTT
e = )(1.02 where A, B, m and n are hydration constants
dependent on the cement type and m alsodepends on Tr (the reference temperature). As
can be seen, the actual reaction time t is
-
7/27/2019 An Cold Thermal
5/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 5
converted to the equivalent time te with the
known reaction temperatures from the heat-
transfer analysis. With this approach, the
hydration and heat-transfer process are coupled
implicitly. The rate of heat generation can beobtained numerically from the above equation
within a given time increment. Typical values
for hydration constants were developed
[VITHARANA, 1995c], and for normal
Portland cement: A=12.56, B=328.7 (both in
kJ/kg of cement) and n=0.42 and m=-1.029 for
the reference temperature Tr= 20oC.
Fig. 3a shows the adiabatic and isothermal heat
generation characteristics for different placing
temperatures for the same concrete mix. The
heat generation within a concrete mass lies inbetween these two extreme conditions. In order
to highlight the significance of the inaccuracy
in using adiabatic hydration models, Fig 3b
shows the typical thermal stress development in
a 200mm and 600mm thick reinforced concrete
walls in moderate ambient conditions
[VITHARANA, 1995b], an overestimation of
the tensile stresses by about 40% .
Fig. 3a Adiabatic and isothermal heat generation
Fig. 3b Thermal stress development with adiabatic
and non-adiabatic models
A general disadvantage with this model is the
difficulty in conducting isothermal heat-of-
solution tests. However, adiabatic tests are easy
to perform and test results are widely available
for different placing temperatures [JCI,ACI207]. Therefore, an indirect method was
developed to synthesise the hydration curves
for varying temperature curves [VITHARANA,
1995b]. This method ignores the temperature-
time history as a direct variable, but the rate of
heat generation is related to the cumulative heat
already generated by adjusting the placing
temperature with the known reaction
temperature at a given time. This was shown to
provide accurate temperature predictions for
non-adiabatic environments.
4 HEAT-TRANSFER
ANALYSIS
Both finite difference and finite element
methods are well established and standard
analysis procedures are available. Finite
difference methods can be formulated easily
and can be implemented even in an Excel
spreadsheet. It is very important to consider the
heat transfer boundaries, particularly the wind
speed and direct solar radiation. The insulationeffect provided by wood formwork should be
considered and sudden removal can generate
surface thermal stresses significant enough to
cause surface cracking. Due to space
limitation, these will not be discussed in detail
here.
Schmidts method is used [ACI207] by design
and construction engineers to predict
temperature rise in concrete placements.
Although this method is useful for preliminary
estimates of thermal stresses, there are severalimplicit assumptions that do not allow an
accurate assessment of temperatures. It
assumes that the surface temperature is equal to
the ambient temperature and field
measurements show that this is not true due to
heat-transfer resistance along the boundaries.
In addition, it uses adiabatic hydration models.
5 MATERIAL PARAMETERS
The development of material and strength
properties in young concrete depends not onlyon the age (time since casting concrete) but also
on the reaction temperature which varies with
-
7/27/2019 An Cold Thermal
6/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 6
time as well as within the structural element
due to the transient heat-transfer. Although the
age has been considered using measured data or
maturity functions [ACI209, CEB] in recent
thermal stress predictions, the influence ofreaction temperature has received less attention
although strength development is also a
thermally-activated process.
5.1 TENSILE STRENGTH AND
YOUNGS MODULUS
The development of tensile strength can be
related to the compressive strength with good
accuracy using the ACI209[1986] strength
development function. The tensile strength ft at
age t (days) at a standard curing temperature of
20 oC is given by:
where ft(28) is the tensile strength at the age of
28 days, and A and B are material constants
dependent of the cement type, curing method,
admixtures etc. Typical values of A and B are:
A=4.0 and B=0.85 for normal-strength concrete
and A=2.30 and B=0.92 high-strength concrete.
The Youngs modulus of elasticity E at age t
can also be related to the corresponding value
of E at 28 days with the above function.
Similar to the heat of hydration, the strength
gain is much faster at elevated temperatures and
this should be considered in the determination
of the thermal cracking index. The major
implication is that the strength gain near the
surface would be much slower compared with
the interior.
The temperature-dependency of ft and E can be
incorporated by using an equivalent (ormaturity) age te which considers the time-
history of the curing temperature. The CEB-
FIP[1991] formulations can be used for this
purpose. The development of compressive and
tensile strength of high-strength concrete in
standard (20 oC) and adiabatic conditions is
shown in Fig. 4. As can be seen, within the
first 3-4 days, the strength gain under adiabatic
condition is about 30% higher than that under
the standard temperature. This highlights the
fact that the purpose of undertaking detailed
thermal stress analysis would be lost if the
development of basic material properties is not
considered rationally.
Fig. 4 Compressive and tensile strength
development under adiabatic and constant
temperatures
5.2 CREEP BEHAVIOUR
The analysis of thermal stresses due to heat-of-
hydration should be carried out in an
incremental fashion in time steps (Section 6).
With known thermal strain developed during a
particular time step, the stress can be calculated
with the appropriate Youngs modulus (Section
3). However, the time-dependent deformational
behaviour of concrete should be considered as
this reduces the magnitude of thermal stresses,usually known as creep-relaxation. In the
analysis procedure developed in THERMAL,
the principle of superposition is used to take
into account the stress-history. Separate creep
factors and characteristics are developed for the
incremental stress developed at each point in
the concrete section during each time step.
It is also important to consider effect of
temperature, before and after a particular
thermal stress is generated, on the creep
behaviour of concrete. The elevatedtemperature generated, before stressing,
determines the effective age te (calculated in
Section 3) thus reducing the creep strain. After
stressing, creep is accelerated with elevated
temperature. Based on an experimental
program [VITHARANA, 1995c], it was
concluded that reasonable accuracy can be
obtained by using the formulations given in
CEB-FIP with some modifications to reflect on
the very early-age concrete behaviour. The
total (elastic and creep) strain (t) at age t undera stress applied at age t = to is given by:
)28(
5.0
tt fBtA
tf
+
=
-
7/27/2019 An Cold Thermal
7/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 7
)(
),()( ]1[toc
tototE
+=
where Ec(to) = Youngs modulus of elasticity at
age to (considering maturity due to reactiontemperature prior to to), o = ultimate creepfactor and (t,to) = function defining time-dependency of creep strain since to. The
ultimate creep value is proportional to the factor
k which defines the age at stressing and the
function (t,to) is given by:
]1.0[
12.0
otk
+=
3.0
),(
+= ta
ttot
6 THERMAL STRESS
CALCULATION
As the analysis is to be carried out in time steps
in an incremental fashion, free strains are to be
determined for each element/node before they
are introduced as induced strains to the
structural analysis program, either a FEM or a
simple beam analysis. THERMAL thencalculates the stresses depending on the external
restraints/boundary conditions of the structure.
FEM analysis would be time consuming and
not always necessary. Simple beam-theory
based calculations can be performed to obtain
accurate results (Section 7). The steps involved
in determining thermal stresses are:
Calculate the incremental thermal strainDt for each point within the concretesection developed during time t and t+t
(ie, within time step n): Dt = [T(t+t)-T(t)].Other induced-strains such as drying and
autogenous shrinkage can also be included.
Calculate the incremental creep strainDcr,to occurring within time step n due to aunit stress ( = 1) applied at an age to:
{ })())(
, tottott
toc
otocr
ED + =
Determine the algebraic sum of the
incremental creep strains Dcr fromincremental stresses generated at the 1sttime step to the current step n.
Calculate incremental thermal stress D(t)developed during time t and t+t. In FEManalysis, strains can be used as the input
directly. The total stress (t) at a given
time is given by the algebraic sum of D(t)up to the current time step n.
crtttttct DTTERFD = ++ )()()()(
The factor RF represents the degree ofrestraint on the free induced strain if this is
known for the structure based on previous
elastic analysis. Axial and flexural
restraint factors RF are given in
[JCI(1986), ACI207 (1973)].
The tendency to cracking is determined bycalculating the Thermal Cracking Index
ratio: Tci = (t)/ft , where ft is the availabletensile strength in hardening concrete. The
time-history of this ratio can be prepared
for the critical points in the structure and
then the probability of cracking can be
determined. Probability of cracking vs Tci
is given in JCI(1986) based extensive field
data. This is a very useful tool for
estimating the probability of cracking for
different concrete mix designs particularly
for comparing temperature controlmeasures and the cost involved.
Thermal stresses are generated in concrete
sections even if they are unrestrained externally
or by structural indeterminacy. These
unrestrained stresses are known as primary
stresses and are caused by non-linear
temperature distributions. Secondary stresses
are generated due to the axial and flexural
restraints. It would be worthwhile and
economical to understand the structure
behaviour in simple axial and flexural actions inconjunction with appropriate restraint factors
before undertaking complex finite element
analysis (eg, walls, massive foundations, box-
girder bridges, dams etc). The influence of
degree of restraint on thermal stresses can also
be used to compare different concrete mix
designs [VITHARANA, 1998] before
specifying them for a particular project.
A typical case for a wall is shown in Fig. 5 in
which unrestrained primary stresses are set up
in the vertical direction while restrained(primary and secondary) stresses are set up in
the horizontal direction due to base fixity.
-
7/27/2019 An Cold Thermal
8/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 8
0
10
20
30
40
50
60
0 10 20 30 40 50 60 70 80
Time (days)
Temperature(C)
Base-measured
Base-predicted
Mid-measured
Mid-predicted
These can be calculated based on the usual
plane sections remain plane hypothesis. Similar
approach has been used to predict stresses in
multi-lift foundation construction (Section 7)
founded on different foundation conditions.
Fig. 5 Thermal stress generation in a wall section
7 APPLICATIONS
It is difficult to firm general conclusions /
recommendations on the development of
thermal stresses due to the involvement of
many factors including ambient conditions. In
the previous sections, the major aspects were
outlined with brief discussions on their
influence(s) on the development of thermal
stresses and tendency to cracking.
7.1 TOKACHI BRIDGE BASE
The spread foundation for this 750 m long
bridge was constructed in the winter conditionswith minimum daily temperatures varying
between 20 to 30 oC. With a plan area of
27mx32m and a thickness of 6m, this was the
second largest substructure of this type ever
constructed in Japan [ISHIKAWA, 1994]. The
construction was carried out in 3-lifts of 2m
thick pours with a 14-day gap between each
pour to permit the dissipation of hydration-
induced heat. Artificial surface heating and
surface insulation was undertaken to prevent
high temperature differentials. The cementcontent used is 280 kg/m
3with normal Portland
cement. During construction, temperature and
strains were measured continuously up to 70
days.
THERMAL was used to compare the
temperatures monitored and to predict thermal
stresses including creep relaxation. Inparticular, it was required to determine whether
the time gap between successive pours (which
is significant compared with the available
construction time) helped reducing thermal
stresses generated compared with a single pour.
In an earlier analysis [ISHIKAWA, 1994], the
measured strains have been converted to
thermal stresses using an effective modulus to
account for the creep relaxation and therefore it
may not be accurate as it neglects early-age
non-linear creep.The temperature predicted by THERMAL is in
close agreement with those measured, Fig. 6.
In particular, the thermal stresses predicted by
THERMAL highlight the importance of
considering the creep-relaxation. The tensile
strength development included the effect of
varying reaction temperatures during the
hardening process. It is interesting to note that
1-dimensional heat transfer and beam analyses
were also undertaken (with zero axial restraint
on the axial deformation) and the predictionswere accurate within 5-10% compared with
finite element analysis results. Therefore, it is
strongly recommended that simple analysis
procedures be undertaken, wherever possible,
which may take a fraction of the time and
resources required for finite element analysis.
Fig. 6a Tockachi Bridge: Temperature in 6 m
thick base
-
7/27/2019 An Cold Thermal
9/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 9
-3
-2
-1
0
1
2
3
4
0 10 20 30 40Time (days)T
ensilestress(MPa)
Base-measured
Base-predicted
Mid-measured
Mid-predicted
Tensile strength
Fig. 6b Tockachi Bridge: Stress generation in 6 m
thick base
7.2 IS MULTI-LIFT
CONSTRUCTION ALWAYS
HELPFUL ?
It is generally believed that by lowering the
average temperature rise, the tendency to
cracking can be minimised. One such method
adopted in concreting for thick foundations is
multi-lift (or multi-layer) concrete placement
with a time gap between each lift [GYRAX,
1994]. A range of parameters was consideredfor foundation thicknesses varying from 1-5m
for typical Hokkaido weather conditions
[VITHARANA, 1995a].
Fig. 7 shows the temperature and thermal
stresses predicted for single and multi-lift
constructions with a 3-day time gap between
each 1-m thick concrete placement. As can be
seen, although temperature is lower with the
multi-lift construction, the induced thermal
stresses are not lower and single-lift concrete
can be undertaken without an increased risk ofthermal cracking. This is due to the fact that
stiffnesses of layers are different in multi-lift
construction thus developing internal restraints
within the structure. The influence of creep is
significant with a stress reduction of about 45-
50%. Following this study, several single-lift
concrete placements up to 5m thick were
undertaken without any thermal cracking.
Fig. 7 Temperature and stress developments in single and multi-lift foundations
-
7/27/2019 An Cold Thermal
10/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 10
0.0
1.0
2.0
3.0
4.0
0 2 4 6 8 10
Time (days)
Stressandstrength(MPa)
0
10
20
30
40
50
Temperature(C)
Tensile strength (MPa)
Tensile stress (MPa)
Temperature (C)
Measured tem (C)
7.3 CANNING DAM ANCHOR
BLOCK
The anchor block for the Canning dam post-
tensioned cables is about 3 m thick [WARK,
2001]. The concrete mix design was based on
previous similar dam constructions (North
Dandalup and Lower South Dandalup) and
extensive laboratory tests undertaken to
minimise the tendency to alkali-aggregate-
reaction. The selected mix had a unit cement
content of 370 kg/m3
(normal Portland cement
= 190, Fly ash = 150 and Silica fume = 30).
With a high Fly ash content, the strength gain
was slower with a 180-day characteristics
compressive strength of 50 MPa (comparedwith the 28-day strength of 31 MPa).
According to the original construction
programme, concreting was to be done during
the summer months and the anticipated main
issue would be the control temperature rise.
The construction specification therefore placed
stringent conditions: a placing temperature of
20oC and a limit of 30
oC on the maximum
temperature rise in the pour. However, delay to
the construction programme required the
concrete placement to be undertaken during thewinter months and there were concerns that the
construction methodology would be unable to
limit the temperature differential to 20oC as per
the specifications.
THERMAL was used to simulate the
temperature monitored in Block No. 20 and
then to predict thermal stresses and the
tendency to cracking. There were also concerns
that uplifting at the old-new interface may
occur with a high differential temperature
gradient. The results at a point 200mm belowthe top of the block are shown in Fig. 8. As can
be seen, the thermal stresses were well below
the tensile strength. As an additional measure,
a couple of layers of ceiling insulation were
used during cooler nights as an insulation layer.
THERMAL considered the influence of creep-
relaxation and further analyses showed that the
predicted stresses would have been higher than
those shown in Fig. 8 by about 30% if creep
effects had been neglected.
Fig. 8 Temperature, stress and strength
development in Canning Dam anchor block (3 m
thick)
7.4 HARVEY DAM INTAKE TOWERFOUNDATION
The foundation for the Harvey Dam intake
tower has a diameter of 20m and a thickness of
2m. The foundation is founded on rock with
grouted dowels for enhancing its stability under
extreme loads. The foundation is provided with
reinforcement to resist the lateral loadings
induced by earthquakes and minimum steel for
crack control.
The total cementitious content was 370 kg/m3
(normal Portland cement =50%, Fly ash =40%and Silica Fume =10%). Temperature
development was monitored during the
construction at several locations. The
construction was carried out in March 2001
when maximum and minimum ambient
temperatures were 32 and 10oC respectively.
THERMAL was used to develop the heat-of-
hydration and heat-transfer in a 1-dimensional
analysis. The results are shown in Fig. 9 for the
mid-thickness of the foundation, comparing
with measured temperatures and they are inexcellent agreement.
-
7/27/2019 An Cold Thermal
11/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 11
0
10
20
30
40
50
60
0 24 48 72 96 120 144
Time (hrs)
Temperature(oC)
Ambient
Measured
Predicted
Fig. 9 Temperature prediction for Harvey Dam
intake tower foundation (2 m thick)
8 CONCLUSIONS
Heat-of-hydration in hardening concrete would
be severe enough to cause cracking in concrete
placements. If accurate results are to be
obtained, correct modelling of the hydration
process is as important as heat-transfer analysis.
The traditional criterion of limiting the
maximum and differential temperature rise
would not always be correct. The primary aim
should be to determine the thermal stresses and
then compare with the strength gain. Stress-
relaxation due to early-age creep is significant.
As shown with practical applications,
THERMAL can be used to predict the
developments of temperature and thermal
stresses with good accuracy.
9 ACKNOWLEDGEMENTS
The first Author would like to thank the staff at
the Hokkaido Development Bureau, Sapporo,
Japan for all the assistance given during his stay
and Dr K Sakai (Head, Structural Engineering)
and Dr N Sato (Head, Dams Engineering) for
guidance and advice on numerous occasions
and financial support. Mr Jonathan Jensen is
thanked for formatting this paper.
10 REFERENCES
AMERICAN CONCRETE INSTITUTE (ACI),
Committee 207, (1970), Mass concrete for
dams and other massive structures.
ACI Committee 209, (1986), Prediction of
creep, shrinkage, and temperature effects in
concrete structures.
ACI Committee 207, (1973), Effect of restraint,
volume change and reinforcement on cracking
of massive concrete, ACI Journal, July 1973.
CEB-FIP, (1991), Model code for concrete
structures, Paris.
CRICHTON, A.J., et al., (1999), Kinta RCC
dam are over-simplified thermal-structurl
analysis valid, ANCOLD Issue No. 115.
GEO-ENG PTY LTD, (2002), Canning dam
construction report.
GYRAX, A., AND SOVACHINDA, A. (1994),
Predicting hydration temperature rise in a mass
concrete stack foundation, Concrete
International, ACI.
HARADA, S., AND MEAKAWA, K., (1991),
Non-linear coupling analysis of heat conduction
and temperature-dependent hydration of cement(in Japanese), Concrete Library of JSCE.
HOLLAND, J.C., AND KRYSA, A., (1991),
Use of silica-fume concrete to repair abrasion-
erosion damage in the Kinzua dam stilling
basin, ACI, SP 91-40, 1991, pp.841-863.
ISHIKAWA, H., AND KONAGI, N., (1994), A
report on the construction of substructure of the
New Tokachi Bridge Winter Mass Concrete,
International Workshop on Low Temperature
Effects on Concrete, Hokkaido, Japan.JAPAN CONCRETE INSTITUTE, (1986),
Standard specifications for design and
construction of concrete structures, Part 2
(construction), Tokyo, Japan.
RASTRUP, E., (1954), Heat of hydration in
concrete, Magazine of Concrete Research, Vol.
6, No. 17, 1954.
THERMAL, (1995), Evaluation of early-age
behaviour of concrete structures by N Vitharana
and K Sakai, Hokkaido Development Bureau,Sapporo, Japan.
-
7/27/2019 An Cold Thermal
12/12
THERMAL CRACKING IN LARGE CONCRETE PLACEMENTS: THEORY AND APPLICATIONS
ANCOLD 2002 Conference on Dams Page 12
VITHARANA, N., AND PRIESTLEY, M.J.N.,
(1998), Significance of temperature-induced
loadings on concrete cylindrical reservoir walls,
Structural Journal, American Concrete Institute.
VITHARANA, N., AND SAKAI, K., (1995a),Single and multi-stage concrete placements for
large foundations: Hydration-induced thermal
stresses, 5th East Asia-Pacific Conference on
Structural Engineering and Construction,
Brisbane, Australia.
VITHARANA, N., AND SAKAI, K., (1995b),
Design against unfamiliar loading conditions,
International Workshop on Integration of
Material and Structural Design, Tokyo, Japan.
VITHARANA, N., AND SAKAI, K., (1995c),Material and creep properties of young
concrete, Report I, Hokkaido Development
Bureau, Sapporo, Japan.
VITHARANA, N., (1995d), Temperature and
stress developments in high-strength concrete
columns under heat-of-hydration effects,
Concrete95 International Conference,
Brisbane.
VITHARANA, N., (1998), Tendency to
cracking in hardening concrete under different
degrees of restraint, 15th AustralasianConference on Mechanics of Structures and
Materials, Melbourne.
WARK, R., VITHARANA, N., WATERS, J.,
AND SOMERFORD, J., (2000), Dam safety
issues and remedial works at Canning Dam in
Western Australia, ICOLD Conference,
Beijing, Sep 2000.