13
1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department of Civil Engineering, IST, Technical University of Lisbon Abstract With this work it is intended to analyse the thermal action on concrete dams, in particular during the construction phase. Numerical simulation of dam’s thermal behaviour is very useful in taking decisions about the construction process, because only by predicting the influence of each variable on the dam’s temperature it is possible to plan the construction in order to control it. Among the factors that allow controlling the dam’s temperature it is possible to highlight: type of formwork and time it remains in place, lift placement rate (lift height as well as time interval between the placement of lifts) and, finally, concrete temperature control, either by using special cements, by prior cooling concre te’s constituent materials, or by artificial cooling after pouring (post cooling). In order to achieve the goal of this work, the general laws of heat transfer by radiation, convection and conduction are analysed. Climatic actions are evaluated as they affect thermal state of dams, at the same time as concrete’s constituent materials and construction methods do. Regarding climatic factors, functions describing air and water temperatures and also solar radiation are estimated. Climatic actions modelling, material properties and the finite element model’s discretization are evaluated in exploration phase, by comparing model’s results with in situ observed temperature. After validating these parameters, the construction phase is studied, which is the main purpose of this work. Construction phase of the dam is simulated. For this purpose a program was developed, updating the model on every construction date and evaluating existing elements as well as exposed and formwork faces at each stage. Hydration heat of Alqueva’s Dam concrete is estimated, and then the construction phase calculation is performed. The methodology used in the development of this work revealed to be suitable because, despite all the uncertainties characteristic of this type of problem, variables are confirmed in exploration phase, allowing achieving results very close to the ones actually observed in construction phase. Key words: Thermal Analysis; Alqueva’s Dam; Climatic Actions; Solar Radiation; Hydration Heat; Construction Phase. 1. INTRODUCTION A proper dam temperature forecast over time requires, to begin with, the use of proper calculation models. Numerical models, in particular based on the finite elements method, make the temperature distribution and consequent structural effects possible to predict with sufficient accuracy, whenever the adopted discretization has enough quality to analyse every action the structure is subjected to. Therefore, it is essential to adopt an appropriate refinement degree of the mesh in order to obtain representative results for all actions. In addition, the finite element mesh should be designed so that its elements enhance a correct modelling of the constructive process (regarding the layers volume and the contraction joints separation), and also in order to provide representativeness in time, simulating faithfully, through mesh increases, the interval between consecutive lifts, as well as its framework in the calendar of the year. These aspects are fundamental to obtain consistent results with the reality. An accurate model of early ages concrete’s behaviour is also of high relevance in the estimation of the dam’s temperature field. Early ages concrete is modelled as a thermomechanical chemoreactive porous media. In this way, the evolution of the hydration reaction is formulated within the theoretical framework for porous media and the thermodynamic problem to solve is nonlinear in the variables temperature and degree of hydration. Additionally, generated temperatures by the heat of hydration are changed by thermal action of the environment. So, it is concluded that the climate action modelling, characteristic of the dam’s location, is also of extreme importance regarding the quality of this analysis results. Hence, one of the focuses of this study is the implementation of a solar radiation model.

THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

  • Upload
    lytruc

  • View
    227

  • Download
    1

Embed Size (px)

Citation preview

Page 1: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

1

THERMAL ANALYSIS OF CONCRETE DAMS DURING

CONSTRUCTION

Application to Alqueva’s Dam

Eloísa Castilho

Department of Civil Engineering, IST, Technical University of Lisbon

Abstract

With this work it is intended to analyse the thermal action on concrete dams, in particular during the construction

phase. Numerical simulation of dam’s thermal behaviour is very useful in taking decisions about the construction process,

because only by predicting the influence of each variable on the dam’s temperature it is possible to plan the construction in

order to control it. Among the factors that allow controlling the dam’s temperature it is possible to highlight: type of

formwork and time it remains in place, lift placement rate (lift height as well as time interval between the placement of lifts)

and, finally, concrete temperature control, either by using special cements, by prior cooling concrete’s constituent materials,

or by artificial cooling after pouring (post cooling). In order to achieve the goal of this work, the general laws of heat transfer

by radiation, convection and conduction are analysed. Climatic actions are evaluated as they affect thermal state of dams, at

the same time as concrete’s constituent materials and construction methods do. Regarding climatic factors, functions

describing air and water temperatures and also solar radiation are estimated. Climatic actions modelling, material properties

and the finite element model’s discretization are evaluated in exploration phase, by comparing model’s results with in situ

observed temperature. After validating these parameters, the construction phase is studied, which is the main purpose of this

work. Construction phase of the dam is simulated. For this purpose a program was developed, updating the model on every

construction date and evaluating existing elements as well as exposed and formwork faces at each stage. Hydration heat of

Alqueva’s Dam concrete is estimated, and then the construction phase calculation is performed. The methodology used in the

development of this work revealed to be suitable because, despite all the uncertainties characteristic of this type of problem,

variables are confirmed in exploration phase, allowing achieving results very close to the ones actually observed in

construction phase.

Key words: Thermal Analysis; Alqueva’s Dam; Climatic Actions; Solar Radiation; Hydration Heat; Construction Phase.

1. INTRODUCTION

A proper dam temperature forecast over time requires,

to begin with, the use of proper calculation models.

Numerical models, in particular based on the finite

elements method, make the temperature distribution

and consequent structural effects possible to predict

with sufficient accuracy, whenever the adopted

discretization has enough quality to analyse every

action the structure is subjected to. Therefore, it is

essential to adopt an appropriate refinement degree of

the mesh in order to obtain representative results for

all actions.

In addition, the finite element mesh should be

designed so that its elements enhance a correct

modelling of the constructive process (regarding the

layers volume and the contraction joints separation),

and also in order to provide representativeness in time,

simulating faithfully, through mesh increases, the

interval between consecutive lifts, as well as its

framework in the calendar of the year. These aspects

are fundamental to obtain consistent results with the

reality.

An accurate model of early ages concrete’s behaviour

is also of high relevance in the estimation of the dam’s

temperature field. Early age’s concrete is modelled as

a thermomechanical chemoreactive porous media. In

this way, the evolution of the hydration reaction is

formulated within the theoretical framework for

porous media and the thermodynamic problem to

solve is nonlinear in the variables temperature and

degree of hydration.

Additionally, generated temperatures by the heat of

hydration are changed by thermal action of the

environment. So, it is concluded that the climate

action modelling, characteristic of the dam’s location,

is also of extreme importance regarding the quality of

this analysis results. Hence, one of the focuses of this

study is the implementation of a solar radiation model.

Page 2: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

2

2. GENERAL PRINCIPLES OF THE

THERMAL PHENOMENON

The temperature variations of a dam are related to the

thermal environmental actions (such as air and water

temperatures, as well as the effects of solar radiation)

and to internal heat generation of concrete during the

construction process.

2.1. Transmission of heat laws

Heat transfer occurs by conduction, convection and

radiation.

Conduction is a heat transfer mode that takes place in

stationary media as a result of a temperature gradient.

The law of heat conduction, also known as Fourier's

law, states that the time rate of heat transfer through a

material is proportional to the negative gradient in the

temperature and to the area, at right angles to that

gradient, through which the heat is flowing:

n

Tk

A

Qq

(1)

where Q is the heat amount crossing the area A , and

q represents the heat flow in the normal direction n

.

k is the materials’ thermal conductivity.

Heat transmission by convection is associated with

heat exchanges within a fluid, or between the fluid and

a surface in contact. It occurs as a result of the fluid

particles movement and is governed by Newton’s law,

expressing the proportionality between the amount of

heat exchanged by convection through a surface per

unit area and time, and the temperature difference

between the surface and the surrounding fluid

expressed by )( aw TT . A convection coefficient, ch ,

is settled, so that:

)( awc TThq (2)

Thermal radiation is a body’s radiant energy emission

process, whose quantity and quality depend on the

body’s temperature. It can be expressed by:

)( 21 TThq r (3)

where rh represents the radiation coefficient,

considered constant in the present work.

Heat exchanges by convection and thermal radiation

can be grouped in a single expression:

)( 21 TThq t (4)

where the parameter th is known as the total thermal

transmission coefficient.

2.2. Internal heat generation

Hydration of cement is a fairly complex set of

competing chemical reactions of different kinetics and

amplitudes [1]. It is a highly exothermic and thermally

activated reaction. Therefore, its simulation requires a

thermochemical model.

The free water present in the mixture reacts with the

unhydrated cement to form hydrates. The water

diffusion through the hydrates layers may be

considered as the dominant mechanism of the

hydration with respect to the kinetics [1].

According to Cervera [2], kinetics relationship can be

expressed in order to represent the normalized affinity,

A~

, as:

RT

EA

dt

d aexp)(~

(5)

where represents the hydration degree, aE is the

activation energy of the reaction, R the universal

constant of perfect gases (8,314 J/(mol K)) and T the

temperature.

2.3. Heat conduction equation

The general transient governing equation for heat

conduction, considering thermochemical coupling

(heat generation thermally activated) can be expressed

as follows:

t

Tc

dt

dmLG

z

Tk

zy

Tk

yx

Tk

xmzyx

(6)

where dtdmLm / corresponds to the thermochemical

coupling and represents the hydration reaction heat

generation. mL is the latent heat, and dtdm /

indicates the reaction speed.

Boundary conditions for the conduction heat equation

can be of two types:

Dirichlet conditions: TT in T (7)

Page 3: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

3

Neumann conditions: Cn

Tkq

in q

(8)

Boundary condition for surfaces with fixed heat flux

and with heat changes by convection and radiation can

be expressed as:

0)(

azyx TThqn

z

Tkm

y

Tkl

x

Tk

in q

(9)

Time appears as a first-order term. Therefore, only

one initial value needs to be specified for the entire

body, that is:

oTT all over the domain at t = to (10)

3. THERMAL STATE QUANTIFICATION

The variables influencing the thermal behavior of

concrete dams are summarized in Table 1:

Table 1: Concrete dams thermal behavior influencing variables

CONSTRUCTION

METHOD

CHARACTERIZATION

Concrete colocation temperature

Contraction joints spacing

Formwork’s type and its placement

duration

Concreting rate

Temperature control measures

THERMAL

CHARACTERIZATION

OF CONCRETE

(Type of cement and its

dosage; Composition of concrete)

Thermal conductivity

Specific heat

Specific mass

Absorption coefficient

Emission coefficient

Hydration heat

GEOMETRICAL AND

LOCATION

CHARACTERIZATION

Latitude

Solar declination

Face azimuth

Thickness

Face slope

THERMAL

CHARACTERIZATION

OF THE

ENVIRONMENT

Air temperature

Water temperature

Convection coefficient (wind

velocity)

Solar radiation

Reflection coefficient of the

surroundings

3.1. Environment’s heat transfer characterization

In order to estimate the dam’s heat loss by its faces in

contact with the air, due to the effects of radiation and

convection, it is necessary to calculate the Newton's

formula coefficient for convection, and the heat loss

by radiation by the Boltzmann formula.

According to Mendes [3], in forced convection, the

respective coefficient can be expressed as:

VVhc (8,3 <50m/s) (11)

where V is the mean wind velocity.

Silveira [4], shown that, for the values of T

observed in reality, the radiation coefficient is nearly

constant, and has an average value of 5,23 W/(m2K).

3.2. Environment’s conditions simulation

In the usual analysis of dams, it is common to

represent the variation of air temperature over time as

the superposition of a medium temperature with two

harmonic functions, one with an annual period and

another with a daily period:

)'(2cos)'(

)'(365

2cos)'(

do

da

ao

aam

tttT

ttTTtT

(12)

where 't is the time, in days, since the beginning of

the year, mT is the average annual temperature, a

aT is

the annual’s wave semi-amplitude, )'(tT da is the daily

wave’s semi-amplitude, aot is the annual’s wave

phase, and dot is the daily wave’s phase.

Similarly, the reservoir’s water temperature presents

an evolution in time with seasonal nature, with the

particularity that the average temperature, )(yTm ,

semi-amplitude, )(yTa , and phase, )(0 yt , vary with

depth, y . In its modeling, it is simply assumed the

existence of an annual variation, disregarding the daily

variation. In this way, the reservoir’s water

temperature can be expressed as:

)('365

2cos)()()',( yttyTyTtyT oam

(13)

The knowledge of the average wind speed on a given

location is essential to determine the heat transfer

coefficient by forced convection and, consequently,

the total heat transmission coefficient between the

concrete’s surface and the air.

Information about the solar radiation, as well as

exposure conditions of the dam’s surfaces, are

fundamental to determine the temperature of these

surfaces, or, more specifically, their temperature

increase in relation to the air’s temperature. At the

Page 4: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

4

ground level, short wavelength solar radiation (global

radiation), I , is given by the sum of three

components: direct or beam radiation, diffuse

radiation from the atmosphere and radiation reflected

by the ground.

3.3. Hydration heat quantification

Cervera et al. [2] normalized affinity’s expression is

used:

exp),(

~ 0

0k

AkTA (14)

In order to estimate Cervera et al. propose the

following expression, according to which, infinite

hydration degree is related with the cw/ (water-

cement) ratio:

cw

cw

/194,0

/031,1

(15)

Therefore, expression (5) results:

RT

En

k

A

n

k aexpexp0

0

(16)

3.4. Construction Methods

Dam’s construction methods influence the initial

conditions of the structure. It is possible, with special

construction processes, to control the temperature after

casting.

Related factors with the construction methods

affecting the temperature are: the spacing of the joints,

the type of formwork used and the time it remains in

place, concreting cadence (that is, the thickness of the

concrete layers and the time interval between placing)

and the temperature control of concrete (achieved

through the use of special cements, prior cooling of

material composition of the concrete, or by artificial

cooling).

When the surfaces have formwork, the total heat

transmission coefficient needs to be adapted.

According to ETL 1110-2-542 [5] it should be

modified as follows:

tformwork hk

bh

1

1'

(17)

4. SOLAR RADIATION LJGK1997 MODEL

IMPLEMENTATION

LJGK1997 model, presented in [6], is used in GIS

(Geographic Information System) implementations

[7], and it is an application of Liu and Jordan’s model

[8], which allows determining the different

components of the solar radiation (namely, the diffuse

one) on inclined plans. In addition, this model has

application anywhere in the world, allowing rough

estimates of the solar radiation amount, which may be

improved if there is locally measured radiation values

information.

From the calculation point of view, solar radiation is a

prescribed heat flux, depending not only on time but

also on the exposed surface orientation. This surface’s

orientation is defined by its normal vector.

4.1. Geometrical relationships

In the equatorial system, solar declination, δ, is

measured on the star’s hour circle from the equator to

the solar vector:

180))3(cos0201,0)3(sen1712,0

)2(cos3656,0)2(sen1149,0

)(cos758,0)(sen2567,233723,0(

(18)

with )346,79(25,365

2 N

[9], where N is the

day of the year considering a year with 365 days.

Hour angle, t, is measured on the equatorial plane

from the observer’s meridian to the projection of the

solar vector. In the local horizontal system, solar

zenith angle, Z, is the angle measured from the local

zenith to the sun, on the star’s hour circle:

tZ coscoscossensencos (19)

where is the latitude.

Solar altitude angle, s , is the complement of solar

zenith angle [6]. The azimuth, , is the angle defined

by the normal’s surface, counted from South, in the

clockwise movement’s direction. The surface’s

inclination angle, Y, is the angle between the

considered surface and the horizontal. The incidence

angle, , is the angle between the direct radiation

incident on a surface and the normal to that surface:

Page 5: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

5

cossencoscossencos tCtBA (20)

with

sensen

cossensencoscos

coscossensencos

YC

YYB

YYA

(21)

4.2. Solar radiation calculation methods

Direct, diffuse and reflected components of radiation

respond differently to the incidence plan’s slope.

Therefore, a disaggregated estimate is needed.

In methods by integration, typically used in clear-sky

conditions, extra-terrestrial irradiance, 0H , and

optical mass of the air, m , are determined based on

Sun-Earth geometry [7]. The extra-terrestrial

irradiance on a surface normal to the solar vector,

nH0 , is given by:

0IHon (22)

Where is a correction factor, and 0I is the

empirical solar constant: 0I =1367 W/m2 (Allen et al.

1998).

In the LJGK1997 method is determined by

Hofierska & Suri expression [10]:

048869,0

25,365

2cos00344,01 N

(23)

(where N varies from 1 to 365 (366)).

4.3. LJGK1997 model

Solar beam radiation, incident on an inclined surface,

in clear sky conditions, icB , is given by:

BBnic FHB cos0 (24)

where BF is a variable which has the value 1 if the

solar disk is visible and 0 if it is hidden.

B represents the beam’s radiation atmospherical

attenuation. According to Kumar et al. [11] it is

estimated by:

)(56,0 )095,0()65.0( mmB ee (25)

where m is the optical air mass:

ssm sen614))sen614(1229( 5.02 (26)

Diffuse radiation is considered to be isotropic and

directly proportional to the visible fraction of the

celestial hemisphere from a given point. This

proportion is given by the sky-view factor, lskyF . .

Hence, in inclined surfaces, diffuse radiation in

clear-sky conditions is given by:

lskyDnskyi FHD .0,)cos( (27)

where, according to Liu & Jordan [8], atmospherical

attenuation to diffuse radiation, D , is estimated by:

BD 2939,02710,0 (28)

LJGK1997 model uses Gates expression to determine

the sky-view factor:

2

cos1

2cos2

,

YYF lsky

(29)

The hemisphere’s remaining fraction contributes with

reflected radiation by the ground. The reflected

radiation component is sometimes neglected, since it

has a low weight in total, except in conditions of high

albedo [7]. Therefore, in this work, this component is

not considered.

4.4. Adaptation to local real sky conditions

Solar radiation in clear sky conditions determined by

LJGK1997 model does not account for the elevation

of the location, nor for climate or environmental

conditions, among others. Therefore, it is not adjusted

to the real sky conditions, particulars to the effective

dam location.

In this work, the adaption to local real sky conditions

was achieved through the comparison of the

LGJK1997 model’s results with Silveira’s [4] solar

radiation values for global radiation on a horizontal

plane (obtained based on Évora registers made in the

50’s). The results are exposed on Fig. 1 and Fig. 2. In

these figures, LJGK1997 model’s clear-sky results (at

different days) are represented discrete and Silveira’s

real-sky values are continuous.

Page 6: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

6

Fig. 1: LJGK1997 model and Silveira’s Global Radiation on a

horizontal plane

Fig. 2: LJGK1997 model’s Beam and Silveira’s Global Radiation

on a horizontal plane

Analysing this information one can observe that, for a

horizontal surface, the direct solar radiation portion

determined by LJGK1997 model adequately simulates

real sky conditions characteristic of the actual dam’s

location. Thus, in this work, this model was adapted to

Alqueva by not incorporating the diffuse component

of the solar radiation. It is important to mention that

this adjustment is clearly justified on horizontal plane.

However, bearing in mind that the different radiation

components respond differently to the dam’s slope,

these conclusions would have to be corroborated with

solar radiation on inclined plane data.

The LJGK1997 model estimated solar radiation flux,

adapted to real and local conditions (by the exclusive

use of the direct component, icB ), is considered in the

static boundary conditions expressed by eq. (2.40)

directly in the term q .

After the LJGK1997 model’s implementation, it is

possible to realize that, only with the introduction of

the dam’s latitude, it is conceivable, for any part of the

planet, to have a very reasonable estimation of the

solar radiation, which can be improved whenever local

registers of solar radiation are available. Therefore,

this model is considered to be a very useful tool in the

climate actions affecting dams’ modelling.

5. TEMPERATURE VARIATION

CALCULATION DURING ALQUEVA’S

DAM CONSTRUCTION

The aim of this work is related to the study of

Alqueva’s dam construction phase. However, given

the number of variables affecting the temperature of

dams under construction, it is necessary to start the

analysis with the study of the exploration phase, in

order to confirm the climate actions admitted

functions, as well as the materials assumed properties.

5.1. Alqueva’s dam characteristics

Alqueva’s dam is a double-curved arch dam. The

dome, with theoretical maximum height of 96,0m, and

development of 348,0m at the crest and 124,0m on the

valley floor, has a thickness of 33,5m at the base and

7,0m at the top. It consists of 24 blocks, generally with

14,5m, limited by vertical joints.

5.2. Materials’ thermal characteristics

In Table 2 thermal model’s properties are described.

Table 2: Thermal model’s properties.

0

200

400

600

800

1 000

1 200

0,0 0,2 0,4 0,6 0,8 1,0

I ch ;

Ih

(W

m-2

)

cos (Z)

0

200

400

600

800

1 000

1 200

0,0 0,2 0,4 0,6 0,8 1,0

Bch

; I h

(W

m-2

)

cos (Z)

Property Unit Value

FOUNDATION

Specific heat, c

Thermal conductivity, k

Specific mass,

Absorption coefficient, a

[J/(kg K)]

[W/(m K)]

[kg/m3]

[-]

879

4,6

2600

0

CONCRETE

Specific heat, c [J/(kg K)] 920

Thermal conductivity, k [W/(m K)] 2,62

Specific mass, [kg/m3] 2400

Absorption coefficient, a [-] 0,65

Total heat transmission coefficient, ht [W/(m2K)] 20,20

Formwork surfaces transm. coef., h’ [W/(m2K)] 2,02

Hydration degree at t , [-] 0,74

Activation energy, REa / [K] 4000

Normalized chemical affinity )(~A (see section 5.5):

0

k [1/s] 555,1360

k

A0

[-] 0,0015

[-] 5,4749

Latent heat, L [J/m3] 6,289.107

Page 7: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

7

In the total heat transmission coefficient

determination, the convection coefficient is estimated

using expression (11), for a mean wind velocity in

Alqueva of 4,0m/s.

5.3. Environment thermal actions’ simulation

Following expression (12) it is possible to write the air

temperature’s function as:

)'(2cos)'()'()'( 1do

da tttTtTtT

(30)

where:

'

365

2sen'

365

2cos)'(1 tbtaTtT m

(31)

Taking into account the records of the daily maximum

and minimum temperature of the air, to Alqueva’s

dam location, the following results can be obtained:

mT =17,49ºC; a =-7,43ºC; b =-3,34ºC.

The annual’s variation of the daily wave’s amplitude

is represented by:

)'(

365

2cos)'()'(2 a

oaam

da tAAtAtT

(32)

where mA , aaA and

ao represent, respectively, the

annual average amplitude, the semi-amplitude of the

amplitude’s annual wave and the annual wave

amplitude’s phase. By doing:

'

365

2sen'

365

2cos)'( tbtaAtA m

(33)

it is obtained, to Alqueva’s dam location,

mA =10,48ºC; a =-3,99ºC; b =-0,82ºC.

In what the water temperature is concerned, in order to

achieve the greatest possible representation of the

hydrological regime, every thermometers belonging to

the upstream faces records were used. Zhu’s formulas

[12] were used to estimate the evolution of the (13)’s

expression parameters with the depth of the reservoir:

(34)

(35)

(36)

(37)

(38)

yefdy )( (39)

where s

mT is the average annual temperature in the

reservoir’s surface, s

aT is the reservoir surface

temperature’s wave semi-amplitude, b

mT is the average

annual temperature in the reservoir’s bottom, H is the

reservoir’s deep, arot is the air temperature’s wave

phase, )(y is the phase difference of the water

temperature relative to the air temperature and , , ,

d and f are constants.

Alqueva’s reservoir water parameters are described in

Table 3.

Table 3: Alqueva’s reservoir water parameters

Parameter Unit Value

smT [°C] 19,95

saT [°C] 7,55

bmT [°C] 11,49

[-] 0,0632

[-] 0,0434

[-] 0,0609

d [months] 3,313

f [months] 2,00

The radiation action implemented in this work is in

accordance with the LJGK1997 model previously

described. Solar radiation estimation requires the

knowledge of the dams’ latitude and orientation.

Alqueva’s dam has a 38º11 latitude and its axis has an

azimuth of 150º.

5.4. Thermal environment actions validation in

exploration phase

In exploration phase the initial temperature field is

estimated assuming an initial temperature, applying

boundary conditions, and performing the time

integration until stationary (cyclic) behavior is

achieved.

Boundary conditions used in exploration phase were:

upstream surfaces submersed (Fig. 3); bottom nodes

of the foundation with a prescribed temperature of

15ºC (Fig. 4); downstream and top faces of the dam

subjected to the solar radiation’s flux (Fig. 5); every

ys

mm ecTcyT )()(ys

aa eTyT )(

)()( ytyt ar

oo

)1/()( gTgTc s

m

b

m

Heg

Page 8: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

8

exposed surfaces subjected to convection and

radiation heat exchanges (Fig. 6); lateral borders of the

model are considered to be adiabatic boundaries.

Fig. 3 – upstream submersed

surfaces Fig. 4 – bottom of foundation

Fig. 5 – solar radiation’s flux

exposed to surfaces

Fig. 6 – convection and radiation

heat exchanges exposed to surfaces

Several finite element model meshes were tested, with

increasing degree of refinement, in order to ensure the

model’s ability to represent the effect of the daily

wave air’s temperature. A model with 5 elements

differently spaced in thickness revealed to be suitable

(Fig. 7).

Fig. 7 – Exploration phase discretization.

In Fig. 8, the results of the exploration phase calculus

(using PAT_2 program [13]) are shown, for 3

particular nodes near the top of the dam, contrasted

with the instruments placed in the same sections

registers. Analyzing the results it is observed that, due

to the used scale, the effect of the air’s temperature

daily wave is reflected as an increase of the thickness

of the annual wave. It is evident this amplitude’s

decrease as the distance to the exposed face increases.

Fig. 8 – Exploration phase temperatures near the top of the dam

(z=141,0m), downstream, middle and upstream, respectively

5.5. Construction phase modelling

According to Ulm and Coussy’s model [1], early ages’

concrete is modeled as a thermomechanical

chemoreactive porous media.

The problem to solve is nonlinear in temperature and

hydration degree (the determination of the

temperature, as well as the hydration degree, depend

on the variables themselves). Furthermore, the

temperature field generated by the hydration heat is

changed by the thermal action coming from the

environment where the dam is located.

The cement used in the composition of Alqueva´s dam

concrete is type IV and class 32,5, according to ASTM

classification.

Analyzing [14] it is possible to understand that, for the

main composition of Alqueva’s dam concrete, the

average quantity of cement is 160,0kg/m3, and the

water-cement average relationship is 0,49.

Based on the results shown in [15], and considering

the composition of 160,0kg/m3 of cement, it is

possible to determine the values presented in Table 4:

0

10

20

30

40

50

0 50 100 150 200 250 300 350

T(ºC) valores calculadosvalores observadosT média ar

0

10

20

30

40

50

0 50 100 150 200 250 300 350

0

10

20

30

40

50

0 50 100 150 200 250 300 350days since the beginning of the year

Calculated values

Registered values

Average air’s temperature

Page 9: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

9

Table 4: Alqueva’s concrete hydration heat.

Expression (15) allows calculating the final degree of

hydration with the average properties of Alqueva’s

concrete composition. Therefore:

74,049284,0194,0

49284,0031,1

(40)

Using the exponential curve suggested in [16] to

represent the accumulated generated heat, Q , it is

obtained to Alqueva’s dam concrete:

teQ

1.8,12

386,19

(41)

The constant Q represents the final amount of

liberated heat in ideal conditions. According to

Cervera [2]:

0TTC

Qad

(42)

where C is the specific heat of the material (920

J/(kgk)), 0T is the initial temperature of the adiabatic

experiment and ad

T is the final reached temperature.

The amount of released heat is given by:

QQ (43)

From the previous relationship it is obtained:

2,26

74,0

386,19

QQ

QQ

(44)

Cervera relates the hydration degree to the

temperature rise in the adiabatic experiment in the

form:

adad

ad

TT

T

(45)

where ad

T is the measured temperature of concrete

along the experiment.

Assuming an initial temperature of 20ºC, the

expression (42) ad

T =41,07ºC is obtained. It is then

possible to calculate the concrete’s temperature in the

experiment as well as its rate adT .

According to Cervera, the normalized affinity can be

expressed as:

ad

a

ad

ad

RT

E

TT

TA exp~

0

(46)

With this information it is conceivable, through the

analytical expression in Cervera’s model for this

function (expression (14)), to calibrate the properties

of the material which fully characterize the chemical

behavior of the concrete mixture. It is obtained:

0

k=555,14s

-1;

k

A0 =0,0015; =5,475

37J/mE29,6240010002,26 QL (47)

For the dam, in construction phase, a finite element

mesh developed in LNEC is used. This mesh has

layers of 2,0 to 3,0m height, and every block of the

dam is divided in 2 in development. The mesh has 4

elements equally spaced in thickness (Fig. 9).

According to section 5.4, the grid should be more

refined near the exposed faces. However, the current

mesh is already computationally demanding, so it was

chosen to maintain the mentioned grid. The results of

the exposed faces will have to be analyzed in the light

of this information. The difference in the obtained

temperature field of the exposed faces, using both

mentioned meshes, subjected to the exploration phase

actions, is presented in Fig. 10 .

The foundation is discretized in conformance with the

grid of the dam in a model with 18 857 elements and

90 434 nodes (Fig. 11).

A program was developed, in order to build the data

archive to be read by the program PATQ_2 [17]. The

designed program considers the evolution of the mesh

in every concreting or stripping phase, updating the

exposed faces (in blue in Fig. 12) as well as the faces

with formwork (in green), and its correspondent total

heat transmission coefficient. A typical output of this

program is the top figure of Fig. 12, which is

contrasted with a photo taken during the construction

of the dam. In red, assumed adiabatic faces are

represented.

t (days) Q (cal/g) Q (kJ/kg)

0 0,00 0,00

3 3,85 16,12

7 4,31 18,05

28 4,63 19,39

Page 10: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

10

Fig. 9 – Construction phase discretization - detail.

Fig. 10 – Exploration phase temperatures near the top of the dam

(z=141,0m) - downstream.

Fig. 11 – Construction phase discretization – general view.

Fig. 12 – Construction process simulation.

In what the initial thermal field is concerned, for the

dam, it was considered that the concrete’s colocation

temperature is equal to the air’s temperature, except in

the cases where it doesn’t satisfy the 27.2 article of

[18]. Regarding the foundation, the initial temperature

field is calculated using PATQ_2 program and

considering the existence of only one phase, in which

every foundation belonging elements are placed,

during a calculus period of time long enough for the

thermal field to achieve the stationary behavior (10

years).

5.6. Results

In the interpretation of the following results it has to

be noticed that, in 2001, artificial cooling process was

carried on, and this action modelling wasn’t

considered in the present study. In Fig. 13 to Fig. 17,

registered temperature is represented in dots, as the

calculated temperature is represented in a continuous

line.

Typical results achieved near the base of the dam are

shown in Fig. 13 and Fig. 14. It is possible to observe

that, when the next layer is laid, there is a decrease in

temperature, due to the lower temperature of the

newly placed layer, followed by an increase due to the

heat generation originated in this second layer. The

same applies, but lighter, when the remaining layers

are placed. The studied instruments are within a layer

located near the base of the foundation (where the dam

is very thick). This location is revealed by the results,

as these nodes are indifferent to the annual’s air

temperature wave effect. In Fig. 13 it is appreciable

the difference in the assumption of the concrete’s

temperature (5ºC) effect. In Fig. 14 it is visible the

different adopted discretization in relation to the real

construction process effect. In this location, for the

layers placed in February ’99, the finite elements mesh

aggregates 3 concreting layers in a single element.

Therefore, the date when the layers are placed will

happen, in the model, later than in reality.

Consequently, they spend more time cooling than

what really has happened.

Characteristic results achieved in the middle height of

the dam, at half thickness, are shown in Fig. 15.

Regarding the studied instrument it should be noted

that the placement of the layer incorporating the quota

of the instrument occurred on 22-07-99 and the next

layer was placed, in reality, on 09-08-99. However,

according to the model’s discretization, it occurred

0

10

20

30

40

0 50 100 150 200 250 300 350

T(ºC)

days since the beginning of the year

valores calculados-malha fase definitiva

valores calculados-malha fase construtivaExploration phase mesh (Fig. 7) Construction phase mesh (Fig. 9)

Page 11: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

11

only on 23-08-99. Therefore, the fact that the layer has

been exposed longer in the model than in reality, in

the summer, caused a higher temperature reached by

the nodes than the registered one.

Typical results achieved in the top of the dam, at half

thickness, are shown in Fig. 16 and Fig. 17. The

presented curves (both the calculated and the

registered one) reflect the position in height of the

studied nodes. In internal nodes near the dam’s crest

(where its thickness is smaller) the cooling is faster

than in lower sections, once exposed faced are closer

to the referred nodes, as well as since the heat is

dissipated more easily by the exposed surfaces than by

the foundation. Thus, the cooling curve has a slope

greater than lower level located sections do. It is also

evident that, for this dam’s thickness, temperature’s

seasonal variations are experimented by internal

nodes, while, at the bottom of the dam, internal nodes

are not affected by these oscillations. In early 2002,

when the curves corresponding to the observed

temperature decrease significantly relative to the

calculated temperature, it is understood that artificial

cooling has occurred in these quotas. This effect tends

to fade over time (as a result of the proximity to the

exposed faces), and the curves meet again, although

artificial cooling has not been modeled. In Fig. 16, the

initial difference is thought to be related to the

concrete’s colocation temperature, although there is no

available information to validate this assumption. In

what Fig. 17 is concerned, the quantity of cement is

identical in the model and in reality. It is inferred that

all other factors (concreting rate, placement

temperature, etc.) have been modeled similarly to

what happened in reality leading to an answer so

similar to what actually occurred.

Fig. 13 –Near the dam’s base results – G27.

Fig. 14 – Near the dam’s base results – G08.

Fig. 15 – Middle height dam’s results – T19.

Fig. 16 – Dam’s crest results – T49.

Fig. 17 – Dam’s crest results – T59.

5.7. Foundation influence

Regarding the effect of the foundation, it is known

that the modelation of the foundation, in relation to its

consideration as an adiabatic boundary, has effect in

the base nodes, since in the first case the heat flow can

be dissipated to the foundation. Thus, for a node in the

dam’s base, in a central section, calculated

temperatures in models with and without foundation

are represented, respectively, in blue and black in Fig.

18. Shown results support the mentioned above. It is

also found that this effect isn’t propagated in height:

by analyzing the mentioned models (with and without

considering the foundation), at the base instruments’

height (approximately distanced 20,0m from the base),

reported in Fig. 19, it is clear that there is no

difference between modeling the foundation or

considering it as simply adiabatic, since both curves

are coincident. It is concluded that, in these sections,

the heat is dissipated by the exposed surfaces and not

by the foundation.

Fig. 18 – Consideration of the foundation in the dam’s base nodes.

0

20

40

13-08-01 02-10-01 21-11-01 10-01-02 01-03-02 20-04-02

T(ºC)

0

20

40

06-01-99 25-07-99 10-02-00 28-08-00 16-03-01

T(ºC)

0

20

40

22-07-99 30-10-99 07-02-00 17-05-00 25-08-00 03-12-00 13-03-01 21-06-01

T(ºC)

0

20

40

06-11-01 25-05-02 11-12-02 29-06-03 15-01-04

T(ºC)

0

20

40

05-07-00 21-01-01 09-08-01 25-02-02 13-09-02 01-04-03 18-10-03

T(ºC)

0

20

40

16-07-98 01-02-99 20-08-99 07-03-00 23-09-00 11-04-01

T(ºC)

Page 12: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

12

Fig. 19 – Consideration of the foundation near the dam’s base

nodes (approximately 20,0m - G05).

6. CONCLUSIONS

The thermal behavior study of dams is essential since

it allows reflecting on the progress of construction

processes, as well as on the different structural types

of concrete dams to adopt, in order to limit the

temperature and to moderate retraction effects

associated with the curing process and hardening of

concrete. Therefore, it is an important tool in the

assessment of the safety of the dams.

In order to plan efficiently the dam’s temperature

control measures, it is essential that this reflection is

based on reasonable estimations of the dam’s

temperature. After completing this study one came to

the understanding that it is possible, by using the

PATQ_2 [17] calculation program, to get very credible

estimates of the dam’s temperature during its

construction, as long as the most relevant factors are

well reproduced.

First of all, and starting with the obvious, it is

necessary to model the climatic action characteristic of

the dam’s location, as well as the materials properties.

Regarding the concrete, it is extremely important to

know its composition, as well as the type of cement

and its dosage, in order to represent the hydration

curve. It is also important to know the type of used

formwork and also the time it remains placed.

In what the finite elements mesh are concerned, it was

noticed that it is essential that its discretization is

adapted to the lift height, as well as to the spacing of

contraction joints, in order to provide the model the

flexibility to simulate the effective constructive

process. If the mesh is not adequately refined, and

combine multiple lifts in only one element, real

boundary conditions will not be well reproduced,

interfering with the hydration heat dissipation, with

repercussions on the elements temperature. Therefore,

it is essential that the finite element mesh has an

adapted discretization to the constructive phasing, not

only in the elements geometry, but also on its updating

over time (existing elements and exposure conditions,

for each phase of concreting or striking), allowing to

represent the real rhythm of casting.

Also regarding the model, it is necessary to refine the

mesh near the exposed faces, in order to be able to

precisely represent the daily wave air’s temperature.

The concrete’s placement temperature is also relevant

in the calculation of the concrete’s temperature rise.

The effect associated to this initial condition tends to

decrease over time, with intensity depending on the

layers exposure conditions. This observation confirms

the limited efficiency of the constituent concrete

materials’ pre-cooling as a technique of temperature

control.

It was found that the modeling of the foundation or its

consideration as an adiabatic boundary only affects the

dam’s base nodes temperature.

The combination of the correct simulation of all these

factors, as noted, allows obtaining very credible

estimates of concrete’s temperature variations during

construction. Thus, it is possible to rehearse, in the

design stage, different rates of concreting, joint

spacing, materials, or even the use of artificial

refrigeration cooling, in order to adopt, in a

knowledgeable manner, the most efficient and

economic construction process at the level of the

concrete’s temperature control.

7. BIBLIOGRAPHY

[1] F.-J. Ulm and O. Coussy, “Modeling of

thermochemomechanical couplings of concrete at

early ages,” Journal of Engineering Mechanics,

vol. 121(7), pp. 785-794, July 1995.

[2] M. Cervera, J. Oliver and T. Prato, “Thermo-

Chemo-Mechanical Model for Concrete. I:

Hydration and Aging,” Journal of Engineering

Mechanics, vol. 125(9), pp. 1018-1027,

September 1999.

[3] P. Mendes, "Acção Térmica Diferencial em

Pontes de Betão", Msc Thesis, IST, 1989.

[4] A. Silveira, "As variações de temperatura nas

barragens". Tese de Especialista LNEC. Memória

nº177, LNEC, 1961.

0

20

40

10-07-99 26-01-00 13-08-00 01-03-01

T(ºC)

Page 13: THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION · PDF file1 THERMAL ANALYSIS OF CONCRETE DAMS DURING CONSTRUCTION Application to Alqueva’s Dam Eloísa Castilho Department

13

[5] USACE - Engineering and Design - Thermal

studies of mass concrete structures, ETL 1110-2-

542, 1997.

[6] F. Néry and J. Matos, “Terrain Parameters in

solar radiation models,” in Proceedings of the

14th European Colloquium on Theoretical and

Quantitative Geography, Tomar, Portugal, 9 to

13 of September of 2005.

[7] F. Néry, "Análise de conjuntos de dados

geográficos de suporte à modelação ecológica da

distribuição de espécies", Phd Thesis, Instituto

Superior Técnico, 2009.

[8] B. Y. H. Liu and R. C. Jordan, The

interrelationship and characteristic distribution of

direct, diffuse and total solar radiation, 1960.

[9] J. G. Corripio, “Vectorial algebra algorithms for

calculating terrain parameters from DEMs and

solar radiation modelling in mountainous

terrain,” Int. J. Geographical Irformation

Science, vol. 17(1), pp. 1-23, 2003.

[10] J. Hofierka and M. Súri, “The solar radiation

model for Open source GIS: implementation and

applications,” in Proceedings of the Open source

GIS - GRASS users conference 2002, Trento,

Italy, 11-13 September 2002.

[11] L. Kumar, A. K. Skidmore and E. Knowles,

“Modelling topographic variation in solar

radiation in a GIS environment,” Int. J.

Geographical Information Science, vol. 11(5),

pp. 475-497, 1997.

[12] B. Zhu, “Prediction of Water temperature in deep

reservoirs,” Dam Engineering Vol VIII Issue I,

pp. 13-25, 1997.

[13] LNEC, “Análise Térmica de barragens de Betão -

Acções térmicas ambientais,” Report 185/2012-

DBB/NMMF, 2012.

[14] “Caracterização das propriedades reológicas do

betão da barragem de Alqueva,” LNEC - NO, (to

be published).

[15] LNEC, “Apreciação sobre o controlo da

qualidade do betão aplicado na barragem do

Alqueva - relatório final,” Report 106/2003 DM.

[16] L. G. d. Silva, “Caracterização das propriedades

termo-mecânicas do betão nas primeiras idades

para aplicação estrutural,” Msc Thesis, FEUP,

Porto, 2007.

[17] LNEC, “Análise Termoquímica de barragens de

betão - Reacção de hidratação,” LNEC Report (to

be published), 2012.

[18] “Regulamento de betões de ligantes hidráulicos,”

Decreto-Lei nº 445/89 de 30 de Dezembro.

Diário da república nº299/89 série I.