Sei sulla pagina 1di 17

A Critical Review of Thermal Analysis of Concrete Dams

T.C. Meri1, P. Moyo2 and M. Alexander3


1
MSc Student, Department of Civil Engineering, University of Cape Town, 7701 Rondebosch, Cape Town, South
Africa. E-mail: meritonicarlos@yahoo.co.uk
2
Associate Professor, Department of Civil Engineering, University of Cape Town, 7701 Rondebosch, Cape Town,
South Africa. E-mail: Pilate.Moyo@uct.ac.za
3
Professor, Department of Civil Engineering, University of Cape Town, 7701 Rondebosch, Cape Town, South Africa.
E-mail: Mark.Alexander@uct.ac.za

Abstract

This paper presents a critical review of thermal analysis of concrete dams in operation. The adopted
mathematical model describing heat transfer process in a dam and the methods of implementation is
presented. The heat transfer processes occurring at dam surfaces including the environmental
factors that influence the temperature distribution are discussed in details. The strength and
accuracy of each solution technique is critically examined and it is found that Finite Element
Method (FEM) is currently the most preferred technique to carry out thermal analysis of concrete
dams accurately, though the more recent Fourier Series Analysis (FSA) technique tries to overcome
some of the difficulties met in FEM.
Keywords: Thermal Analysis; Fourier Heat Model; Environmental factors; Implementation
Techniques

INTRODUCTION
Thermal analysis is an important aspect of safety and behaviour study of concrete dams.
Transient heat transfer analysis due to variable environmental temperature and the associated
thermal stress computations are the key areas of interest to researchers. Knowledge of thermal
stresses is essential for the development of a defensive measure in dam safety analysis. Although
the basic theory and computational methods have been established, the heat transfer processes
occurring in and around the dam as well as the influence of the environment to temperature
distribution still remain a big challenge to most researchers.
The present review is intended to encompass the thermal analysis of concrete dams in
operation in papers published between 1990 to date. While being exhaustive, some selection is
necessary. Many papers reviewed herein relate to the process of heat transfer, the environmental
factors influencing temperature distribution in concrete dams and the current state-of-the-art
implementation techniques adopted. We have restricted ourselves to papers published in reviewed
archival journals and a few selected concrete dam manuals and heat transfer text books.
FOURIER HEAT CONDUCTION MODEL
In order to accurately obtain the temperature distribution in concrete dams, an appropriate
mathematical model describing the heat flow within the concrete has to be established. Currently,
the heat flow within a dam is assumed to be through conduction only (Leger et al. 1993a; Agullo et
al. 1996; Malla and Weiland 1999) and hence governed by the transient Fourier heat conduction
model. The Fourier heat conduction model in a solid region is generally written in cartesian as:

Q 1 T
2T   [1]
k Dm t

2 2 2 k
Where:   2  2  2 Dm 
2

x y z ρc

2 is the Laplacian operator of temperature, T denotes temperature field T(x, y, z, t) and x, y, z and
t denoting space coordinate vector and time; Dm is the thermal diffusivity of the material; k is the
tensor which quantifies the conductivity property of the material; c and ρ are material specific heat
capacity and density respectively, Q is the internal rate of heat generation in W / m2 . Special cases
of Eq. (1) are;
a) Poisson heat model which describes steady-state heat conduction phenomena
Q
2T   0 [2a]
T
b) Transient Fourier heat conduction model with no internal heat generation term
1 T
2T  [2b]
Dm t
c) Laplace heat model which describes steady-state heat conduction phenomena with no internal
heat generation term

2T  0 [2c]
Eq. (2b) is the adopted mathematical model for heat transfer analysis of concrete dams in
operation (Leger et al. 1993; Agullo et al. 1996; Daoudi et al. 1997; Malla and Weiland 1999;
Sheibany and Ghaemian 2006; Ardito et al. 2008; Labibzadeh et al. 2010). It is assumed that for
concrete structures such as dams that have been in operation for more than 10 years (Nisar et al.
2003, Saetta et al. 1995), the hydration process is complete and therefore internal heat effect due to
cement hydration can be ignored.
Eq. (2b) is normally applied in three different forms in carrying out thermal analysis of
concrete dams: These are; one-dimensional (1-D), two-dimensional (2-D) and three-dimensional (3-
D) transient Fourier heat conduction models. Mathematically these three forms are written
independently in cartesian system as:
1-D model:

2T T
kx  ρc [3a]
x 2
t
2-D model:

2T 2T T
kx 2  ky 2  ρc [3b]
x y t

3-D model:

2T 2T 2T T


kx  k  k  ρc [3c]
x y z t
2 y 2 z 2

Where; T is the concrete temperature in Kelvin, kx , ky andkz are respectively concrete thermal
conductivity coefficients along the x-, y- and z-directions. In 1-D heat transfer analysis, heat flow is
considered only along the thickness (L) of the dam i.e. in the x-direction. Whereas In 2-D heat
transfer case, heat flow is considered along the dam thickness (L) and through the dam depth (H)
i.e. heat flow occurs in both the x-and y-directions. Finally in 3-D heat transfer analysis, heat flow
is considered in all three directions i.e. in the x- , y- and z-directions. In the z-direction, heat flow is
considered across the radius (R) of the dam. Diagrammatic representation is shown in Figure (1)
below

Figure 1: 1-D, 2-D and 3-D heat transfer Cartesian representation


HEAT TRANSFER ANALYSIS
The basis for heat transfer analysis in concrete dams is the governing heat transfer equation
which is based on the principle of energy conservation using classic Fourier’s heat conduction law
(Zhang et al. 1998). Concrete dam scholars often synonymously used the term thermal analysis with
heat transfer analysis/ heat flow analysis/ temperature field analysis to mean obtaining the solution
to Eq. (2b) which means determining the temperature distribution within the dam structure. An
accurate solution to Eq. (2b) can only be obtained through a realistic definition of boundary
conditions at the dam surfaces (Leger et al. 1993).
The heat conduction phenomenon can be accurately applied to approximate the temperature
field within the dam. However there are a number of interrelated mechanisms occurring at the dam
surfaces exposed to the environment. Fig (2) describes generally the heat transfer processes
occurring in a dam-foundation-reservoir system and the boundary conditions usually adopted in the
analysis.

Fig.2: Heat transfer processes and boundary conditions

Heat transfer processes


For concrete dams in operation, heat transfer is mainly governed by three mechanisms:
a) Conduction through the wall
b) Convection to and from the concrete surface by the surrounding fluid
c) Radiation exchange between the concrete surface and the surrounding

Heat conduction
This is the main process controlling heat flow within the concrete wall including the concrete
surface. This occurs as result of temperature gradient within the wall. The amount of heat energy
transferred between two points within the dam at different temperature can be conveniently
obtained using Fourier’s law as

T
q  kA [4]
n

dT
Where: q is the heat flux in W/m2, is the temperature gradient in the direction normal to area A,
dn
k is thermal conductivity for the concrete medium and normally dependent on the concrete
temperature

Heat convection
The concrete dam is exposed to two different moving fluids at its surface that is water on the
upstream face and air on the downstream face. Heat energy transfer occurs to and from the concrete
surface whenever a temperature difference exists between the media and the concrete. If the
temperature of the fluid /surrounding medium is Ts and the concrete surface temperature is T, the
heat transfer per unit time is given by Newton’s law of cooling as

q  hc A T  Ts  [5]

Where: hc is convective heat transfer coefficient which defines the constant of proportionality
relating the heat transfer per unit time per unit area A to the overall temperature difference. It must
however be noted the fundamental energy exchange at the concrete boundary is by conduction and
that this energy is carried away by the fluid flow

Heat radiation
Whereas conduction and convection modes of heat transfer are proportional to linear
temperature difference, experimental evidence indicates that radiant heat transferred is proportional
to the fourth power of the absolute temperature. This is process is governed by Stefan Boltzmann
law given as

q  eCA T 4  Ts4  [6]

Where: e is the emissivity of the concrete surface; C is Stefan-Boltzmann constant independent of


surface, surrounding and their temperature. Its value is 5.6697 X 10-8 W/ (m2 .K4)
Eq. (6) as applied to thermal analysis of concrete is usually written in form

q  hr T  Ts  [7]

Where: hr is defined as linearized radiation coefficient (W/m2) given as

hr  eC T 2  Ts2  T  Ts  [8]
Boundary conditions
The treatment of boundary conditions in heat transfer analysis is the most complex.
Appropriate and accurately defined boundary conditions at dam surfaces will certainly lead to
reliable results. The boundary of a dam consists of dam-air boundary, dam-water boundary and
dam-foundation boundary.

Concrete-air boundary
The dam-air boundary is always a difficult problem to deal with. A large difference exists
between the temperature of the air and concrete. The heat transfer mechanisms at this boundary are
mainly due to radiation and convection. The heat exchange phenomena here are very complicated.
This is because so many factors need to be concerned with such as air temperature, wind speed,
solar radiation and light reflection. Difficulties may arise in the determination of solar radiation
intensity as well as the convective coefficient, hc. Because hc is a function of air flow, the
convective coefficient may be environment dependent.

Air temperature
The daily and seasonal air temperature variation is largely due to the horizontal movement of
air, resulting from movement of relatively cold or warm air masses. Other factors which can greatly
influence air temperature are physical processes such as absorption and emission of radiation, heat
conduction near concrete surface.
Estimates of the air temperatures at a dam sites are made based on recorded data. Researchers
have observed that, air temperature at the dam site varies in an almost sinusoidal fashion over given
period of time (Leger et al. 1993a; Agullo et al 1996; Sheibany and Ghaemian 2006) and hence can
be approximated as:

 2π(t  ξ ) 
Ta  A sin   Bm [9]
 365 

A = amplitude in °C; Bm = annual mean temperature in °C; t = time in days; ξ = lag factor of
sinusoidal representation of daily temperature
The amplitude of the sine wave, A, can be determined by the following

A  0.5 Tmax  Tmean  Tmin  Tmean  [10]

Tmax = maximum average monthly temperature, Tmin = minimum average monthly temperature,
Tmean = yearly mean average temperature

Heat transfer coefficients


The coefficient of heat transfer by convection hc on this boundary is usually treated as an
input data. For this boundary Eq. (5) can be rewritten as

q  ha T  Ta  [11]

Where: ha is convective heat transfer coefficient at air-concrete boundary


This coefficient, which depends on many factors is a function of primarily wind velocity and can be
related to it empirically as proposed by several researchers as:

ha  3.83V + 4.67 (Kehlbeck, 1975) [12a]

ha  3.8V + 5.7 (Kreith and Kreider, 1981) [12b]

ha  3V + 2.8 (Duffie and Beckman, 1980) [12c]

Where V is the wind speed at the dam location in m/s


Similarly the heat transfer coefficient due to radiation is defined as a function of air temperature and
the concrete surface temperature as

q  hr T  Ta  [13]

Where hr becomes

hr  eC T 2  Ta2  T  Ta  [14]

Solar Radiation on Concrete Faces


Solar radiation energy incident on the concrete surface is either reflected or absorbed by the
surface. The amount of solar energy absorbed by the concrete surface can be obtained as

q  aH [15]

Where: H is total amount of incident solar energy per unit area per unit time, and a, is the
absorptivity of the concrete surface.
However in the absence of actual measurements of the local solar radiation, the data that is
commonly used pertain to the monthly average of the daily global solar radiation on a horizontal
surface (Ho) (Agullo et al. 1996). Using the monthly average (Ho), the method proposed by Liu and
Jordan (1967) allows the calculation of the amount of daily sky diffused radiation (Hd) by obtaining
the direct component of radiation, (Hb) as the difference between the two as

Hb  Ho  Hd [16]

The relation between Ho and Hd is mathematically expressed as

Hd  Ho 1.39  4.027KT  5.531KT2  3.10KT3  [17]

Where: KT is the index of the average monthly cloudiness, defined by the ratio between H o and the
monthly average of extraterrestrial solar radiation He and this can be evaluated as
Ho π
KT  [18]
24r 2 .Isc  cos δ cos φ sin hs   hs sinδ cos φ 

Where, r2 is the correction factor of the solar constant for every day of the year
 360 Z 
r 2  1  0.003cos   ,1  Z  365 [19]
 365 

Where: Isc is solar constant with value of 4870.8 KJ/hm2; δ is the latitude of the location; φ is the
solar declination and hs is the absolute value of the hourly angle corresponding to sunset, expressed
in radians
The declination can be obtained from tables or approximate formulae (Labibzedah et al. 2010)
that express the declination as a function of the day of the year for example Duffie and Beckman
(1974). Nevertheless, a representative day of every month is usually obtained in order to calculate
the declination. This day is usually taken to be that when the extraterrestrial radiation sis closest to
value of the average daily extraterrestrial radiation during that given month. Coronas et al. (1992)
presented a representative day for every month and the corresponding value of the solar radiation as
given Table 1 below
Table 1: middle days and their solar declinations

Month Middle degree


day
January 17 -20.7
February 15 -12.6
March 16 -1.7
April 15 9.8
May 15 18.9
June 10 23.0
July 17 21.2
August 17 13.4
September 16 2.6
October 16 -8.9
November 15 -18.5
December 11 -23.0

The hourly angle (hs) corresponding to the sunset is obtained from

cos  hs    tanφ.tanδ [20]


Where -hs is hourly angle that corresponds to sunrise
Knowing the hourly angles of sunrise and sunset, the duration of the solar day (TSV) can be
determined. This duration is the time between two consecutive passes of the sun over the longitude
of the location. The relation between TSV (in hours) and the hourly angle (in degrees) is given by

hs  15(TSV  12) [21]

Consequently, the beginning of solar day (TSVi) and the end of solar day (TSVe) are given by the
relations below:
TSVi  12 
1
15
 cos1 ( tan φ.tan δ)
[22]

TSVe  12 
1
15
 cos1 ( tan φ.tan δ) 
[23]

TSV  12 
2
15
 cos1 ( tan φ.tan δ)
[24]

The above relations (22-24) depend on the latitude φ of dam location and the day of the year (Z).
Thus for a given dam the interval of the solar radiations at its location can be determined. Outside
this interval, the incident solar radiation is taken to be zero.
Once the direct and diffuse components of the monthly average of the daily global radiation at
the location (horizontal plane) have been determined, we obtain the hourly radiation at different
times during the interval that corresponds to sunrise and sunset at the location of the dam. More on
hourly radiation can be obtained from Agullo et al. (1996).
From the direct and diffuse components of the solar radiation, the amount of incident solar radiation
on the dam surface is computed as the sum of their hourly beam, diffuse and reflected components

H  Hb ,h  Hd ,h  Hr ,h [25]

Where: ground reflected radiation Hr ,h is given as

 1  cos β 
Hr ,h  p  .(Hd ,h  Hb ,h ) [26]
 2 

β is the sloping angle of dam surface

Concrete-water boundary
The situation at the water-concrete boundary is similar to that of air-concrete boundary. It is
normally assumed that the whole upstream face of the dam is completely covered with water to
simplify the analysis process. This assumption is well represented in almost all the papers reviewed
here in. This assumption is very practical and acceptable given the fact that when there is no water
on this face then the situation as describe in air-concrete case can be applied.
Heat exchanges convectively between the dam and the water. For simplicity, Leger et al.
(1993) suggested that, for this boundary, the concrete in contact with water can be assumed to have
the same temperature as the water temperature. This means an error rather negligible is introduced
at this boundary during the analysis. Nevertheless this has been adopted by several researchers
(Agullo et al. 1996; Daoudi et al 1997; Malla and Weiland 1999; Sheibany and Ghaemian 2006,
Labibzedah et al. 2010).

Water temperature
Temperature of the impounded water is a function of time and depth of the reservoir. There is
no general rule to define reservoir temperature profile. Closed-form expressions to obtain the
reservoir temperature profile and its variation with time have been suggested by Bofang and
Zhammei (1997) and Ardito et al. (2008).
The proposed empirical formula by Bofang and Zhammei provides the water temperature Tw
as a function of time and water depth yw below the reservoir level. This is expressed as:

Tw (yw ,t)  Tm (y)  Au (y)cos ω(t  to  ξ ) [27]

Where: t is time in days; yw is the water depth; Tm is the annual mean temperature of water obtained
as:

Tm  co   b  co  e0.04 yw [28]

Tb  bg
Where; co  ; g  e0.04H Tb is the water temperature at the bottom; b is the annual mean
1g
water temperature at the surface of the reservoir; H is the depth of the reservoir; Au (yw ) is the
amplitude of annual variation of water temperature given by

Au (yw )  A0e 0.018 yw [29]

Where: A0 is the amplitude of annual variation of water temperature at the surface of the reservoir;
t o is the time of maximum air temperature; ξ is the phase difference between the maximum
temperatures of water and air.
Eq. (27) was slightly modified by Ardito et al. (2008) which make use of measured water
temperature. The modified form is given as:

1  e Φyw e Φyw  e ΦH


Tw  yw ,t   Tb (t )  T (t ) [30]
1  e ΦH 1  e ΦH
t

Where Tb and Tt are time sequence of temperature measurements at yw = H and yw = 0 respectively


and Φ represents an empirical parameter which must be evaluated for the dam under investigation
(for instance Bofang and Zhammei proposed Φ  0.04 for their situation)
Zhang (1998) however decided to use an equivalent heat flow defined in Eq. (5) to prescribe this
boundary. In his case the heat convection model in Eq. (5) is written as

q  hw T  Tw  [31]

Where: hw is the convective heat coefficient for water-concrete boundary. Several values of hw and
ha were tested in analysis and a good agreement between field measurements and computed
temperature was obtained for hw = 20 W/m2 K and ha = 7 W/m2 K. It was also observed that by
assuming a very large hw makes the temperature at the water-concrete boundary equal to water
temperature. Hence the earlier assumption by Leger et al. (1993) can be considered appropriate for
analysis based on the fact these heat coefficients vary throughout the year and from one dam site to
another.
Concrete-foundation boundary
The discussion relating to this boundary is not thorough in literature. At the dam-foundation
boundary, there seem to be a lot of uncertainty about what actually occurs. In some cases a
simplified adiabatic condition is assumed at the interface. The reason being since concrete is a fair
insulator; this kind of simplification does not influence significantly the thermal response away
from the interface (Leger et. al, 1993a). In other cases, either temperature or heat flux is prescribed.
The prescribed temperature condition has recently been accepted as the most appropriate (Ardito et
al. 2008) considering thickness of the dam wall.

Implementation of heat transfer processes and boundary conditions


The complete solution of the heat balance equation in (2b) in order to obtain the temperature
distribution in concrete dams employs the use of sophisticated numerical techniques such as finite
element method (FEM), finite difference method (FDM), finite volume method (FVM) and Fourier
series analysis (FSA) in order to establish the temperature field in the dam.
The finite element analysis (FEA) technique besides its limitations such as large
computational effort and required experts experience is currently the most preferred approach to
handle this task. This is particularly due to its ability to deal with irregular geometries and complex
boundary conditions and loadings which are the biggest limitations of FDM and FVM. Fourier
series analysis on the other hand comes with some advantages over FEM like less computing time
requirements and no need for initial temperature condition (Ardito et al. 2008) which is an
important requirement when using FEA and FDM.

Finite Element Method (FEM)


The methodology in assessing seasonal temperature variation in gravity dams by finite
element analysis has been described by Leger et al. (1993a). The thermal behavior and subsequent
structural response of typical concrete dams can be represented accurately by 2-D and 3-D finite
element models. It must be however noted that generally concrete dams are 3-D structures from a
stress stand-point and a 2-D from heat-flow stand-point as little heat is transmitted in the direction
normal to the vertical plane i.e. longitudinal through the dam (U.S Army Corps, 1995). Thus in the
case where stress response is depended on the thermal output then it is usually in the best interest to
use 3-D models in the heat transfer analysis to avoid creating a structural model again.
In the finite element solution technique, three major computational issues ought to be treated
properly. Firstly, is the initial condition of the dams, the second is related to the numerical
integration scheme when a long time behavior is required and finally the boundary conditions.
These are fundamental factors in carrying out thermal analysis using finite element analysis
For an arbitrary 3-D body depicted in Figure (3), the governing equation for transient heat transfer
is given in Eq. (2c)
Figure 3: A Body subjected to heat transfer
On the surface of the body, the following conditions must be satisfied:

T  x , y , z ,t   Tini  x , y , z ,t  on ΓT [32a]

T
κ  qs  x , y , z , t  on Γq [32b]
n
Where: qs is heat flux, ΓT is temperature imposed boundary part and Γq is heat flux boundary part.
There are three kinds of contribution to the heat flux at the surface. A general expression that
describes all the effects is

qs  qa  qc  qr [33]

Where: qa , qc , qr are; heat flux due to absorbed solar radiation, convection and surface radiation
respectively discussed earlier.
The initial condition given by

T  x , y , z ,0   Tini  x , y , z  [34]

This is prerequisite for the solution of the initial-boundary-value problem given by equations (2c,
32 – 33)
The finite element formulation of Eq. (2c) can be derived using Galerkin method based on the
variational principle. The integral form (often called the weak form) in 3-D is written as

 .

  w  DTdΩ    w  ρcT  Q dΩ    wq ndΓ  0 w
T
T
with w  0 on ΓT [35]
Ω Ω Γq
,

Where Q = heat or thermal source

on

L = dam thickness, w = test or weight function, D = generalized thermal conductivity


For the temperature T within the element, an approximation can be written in the form

T  Nd [36]

Where N is the array of shape functions and d is the vector of nodal temperatures
After several manipulations of the integral form, the resulting algebraic equation for each element is
an ordinary differential equation in time written as

CT t   KT t   F t  [37]

If m denotes the number of nodes in the adopted finite element mesh, C and K are the (m x m)
matrices of thermal capacity and (thermal conductivity + convective + radiative) respectively that is

K  K1  K2  K3 [38]

T and F denote the m-vectors of unknown nodal temperatures and nodal force respectively, the
latter of which reads in integral form as:

F t    NT QdΩ   NT  qT n  dΓ [39]
Ω Γq

At the element level

C   ρcNT NdV ; K1   kBT BdΩ; K2   hc NT Nds; K3   hr NT Nds [40]


Ωe Ωe Γe Γe

F   QeNT dΩ   qaNT ds   hcTaNT ds   hrTaNT ds [41]


Ωe Γe Γe Γe

Where B is the derivative of the shape functions defined as

 NT NT NT 


B 
T
 [42]
 x y z 

Initial condition
The initial condition is necessary for the transient temperature field analysis of a dam
structure in finite element analysis. Usually the initial data is not available unless the time history of
dam is known. The common method for prescribing the initial temperature field is to assume a
certain type of steady temperature solution as the initial condition (Venturelli 1992, Leger et al.
1993).
To start transient computations, a steady-state heat transfer analysis is carried out by applying
the mean annual air, water and foundation temperatures directly at the boundaries of the model. The
mean annual temperature on the downstream face is increased above the mean annual air
temperature to account for the effect of solar radiation. This is done for the extreme temperature
conditions usually the year with maximum recorded temperature (Leger et al. 1993). Boundary
temperature variations of certain period are repeated in time. When a repetitive temperature time
history of a 2 to 3 year simulation is observed, the transient heat transfer analysis is considered to be
convergent. Only the results after convergence is achieved seem reliable for practical use. A finite
period of the time is required beyond which errors introduced by the assumed initial field
temperature to drop to a negligible level.

Boundary conditions
The boundary of a dam consists of dam-air boundary, dam-water boundary and dam-
foundation boundary. The basic information required to perform a complete heat flow analysis
using finite element analysis include ; 1) weather data describing the air temperature, solar
radiation, wind speed; 2) water temperature 3) foundation temperature; and finally 4) thermal
properties of the materials. Suitable computational models have been established to treat these
factors as already discussed for most cases. This helps during the implementation in finite element
software packages.

Time step solution of the FEM equations


Once the initial and boundary conditions have been established, the solutions to finite element
Eqs. (37-42) can be obtained. There exist mainly two methods: the mode superposition method and
the direct integration method. The method of choice is usually based on whether the long term or
short-term solution is required and the scale of the problem.
In order to integrate Eq. (37) with respect to time, the time interval is first divided into equal steps.
This is written as:

to  0 and ti  it , i  1,.........., n. [43]

Where t is the time step length and tn  nt is the final time of interest and n is the maximum
number of steps. The time derivative of the temperature array, T , at ti 1 is approximated by the
backward difference scheme, i.e.

1
Ti 1  Ti 1  Ti  [44]
t
Substituting Eq. (44) into equation (37), for time ti 1 , we obtain the following difference equation:

C  tK  Ti 1  CTi  tFi 1 [45]

For all the time steps, the matrix C  tK  is the same, so that the marching scheme is
CTi  tFi 1 
1
Ti 1  C  tK  [46]

Eq. (46) therefore can be used to determine Ti 1 in terms of Ti. To is actually the initial temperature
distribution array Tini . The transient behavior of the concrete can then be determined step by step.
A few research works on thermal analysis studies of concrete dams in operation using finite
element analysis technique have been carried out (Leger et al. 1993a and b; Daoudi et al. 1997;
Malla and Weiland 1999; Manafpour et al. 2000; Nissar et al. 2006; Sheibany and Ghaemian 2006;
Labibzedah et al. 2010). Several assumptions were adopted by different researchers for the different
cases studied. These were dependent on the objective to be achieved and due to the fact that
environmental parameters varied from location to location.

Finite Difference Method (FDM)


Finite difference method was used to obtain the temperature distribution in the past
particularly in the 1-D heat transfer analysis (Agullo et al. 1996). The procedure involved is very
similar to finite element method but only differs from the way it generates algebraic equations. The
core idea behind FDM is that the derivatives in Eq. (2c) are replaced with the difference formulas
(Chapra and Canale, Numerical Methods for Engineers, third edition provides in details the
explicit/implicit FDM approach).
The fundamental assumption in 1-D heat transfer analysis is the unidirectional flow of heat which is
usually not true especially for concrete dams with considerable thickness. This makes the 1-D model in most
cases not considered strictly accurate in concrete dams due to the fact that in reality there may be heat flow
normal to the x-direction.
In addition FDM requires the use of structured grids for geometric discretization which affects the
accuracy of the final result especially when dealing with spherical or curved surfaces. Besides the errors
obtained as result of conversion of the differential form of the governing equation to simple formulations that
degrades the accuracy of the numerical solution, the method also makes it difficult to model domains with
complex geometries and irregular boundaries (Sabbagh-Yazdi and Mastorakis, 2007; Janna, 1988)

Fourier Series Analysis (FSA)


This technique is a new approach proposed by a few dam researchers (Leger and Leclerc
2007; Ardito et al. 2008). Leger and Leclerc proposed approach involves the solution of direct and
inverse thermal problems using Fourier series functions assuming periodic thermal excitations for
harmonic and non-harmonic thermal signals. This approach requires temperature measurements
inside the dam for the inverse analysis which might be a limitation in cases without embedded
thermo couples inside the structure. Whereas Ardito’s proposed algorithm used the FEM
discretization process to reduce 2-D heat model to first order linear differential equation. Then using
Fourier series functions, an analytical solution of the Ordinary Differential Equation (ODE) was
obtained that provided the temperature field over given time period, assuming steady harmonic
thermal response of the dam. Full scale assessment of the validity of this approach using real
experimental data is yet to be done. Leger and Leclerc concluded that the accuracy of computed
thermal field is independent of the mesh distribution of the section considered which is in contrast
to finite element heat transfer solutions. On the other hand Ardito concluded that in their proposed
technique, initial conditions are not required.
SUMMARY

Temperature distribution in the concrete dams in operation is controlled by seasonal


temperature and climatic conditions. The seasonal variations in air, reservoir and foundation
temperature as well as solar radiation including wind effects are responsible for temperature field in
concrete dams in operational stages.
Knowledge of concrete temperature condition is essential for the analysis of thermally
induced stresses within the structure and therefore forms the basis of thermally induced stress
computations.
A realistic and reliable thermal analysis methods and computational models have been well
established. Fourier heat conduction model is presently adopted as the mathematical model to
define the heat transfer in concrete dams. The FEM is currently the most preferred solution
technique to implement the heat transfer processes and boundary conditions in order the accurately
determine the temperature distribution in concrete dams in operation.
Parametric models for ease of implementation in FEM packages have been established that is
air temperature model (Leger et al. 1993) and water temperature model (Bofang 1997; Ardito et al
2008). The foundation temperature model is however lacking.
The effects of air, water temperature, solar radiation and thermal properties have been
adequately dealt with (Leger et al. 1993; Agullo et al. 1996; Daoudi et al. 1997, Malla and Weiland
1999; Sheibany and Ghaemian 2006). We have observed that the effect of foundation temperature
to overall dam temperature distribution has not been thoroughly dealt with in literature. A detailed
understanding of the influence of foundation temperature to concrete dam temperature distribution
follows in the sister paper.

CONCLUSION

Accurate prediction of concrete dam’s temperature distribution requires a good understanding of


thermal interaction of the dam and its environment. Realistic assumptions and appropriate
parametric models including accurate parametric data are essential for finite element
implementation in order to obtain reliable results

REFERENCES

[1] Daoudi, M., Galanis, N., and Ballivy, G. (1997). “Calculations of the Periodic temperature
field in concrete dams.” Can. J. Civ. Eng., 24(5), 772-784.
[2] Leger, P., Venturelli, J., Bhattacharjee, S.S. (1993a,b). “Seasonal temperature and stress
distribution in concrete gravity dams.” Can. J. Civ. Eng., 20(6), 999-1017.
[3] Malla, S.,Wieland, M. (1999). “Analysis of an arch-gravity dam with a horizontal crack”
Comput Struct, 72, 267–78.
[4] Ardito, R., Maier, G., and Massalongo, G. (2008) “Diagnostic analysis of concrete dams based
on seasonal hydrostatic loading.” J. Mech Behaviour Mater, 15(6): 381–9
[5] U.S. Army Corps of Engineering.(1994). “Arch dam design.” Engineer manual 110-2-220,
Chap. 8, 1–15.
[6] Sheibany, F. and Ghaemian, M. (2006). “Effects of environmental action on thermal stress
analysis of Karaj concrete arch dams” J. Eng. Mech., 132(5), 532-544.
[7] Bofang, Z. (1997). “Prediction of water temperature in deep reservoir” Dam Eng., VIII (1), 13–
25
[8] Agullo, L., Mirambell, E., and Aquado, A. (1996). “A model for the analysis of concrete dams
due to environmental thermal effects” Int. J. Numer. Methods heat fluid flow, 6(4), 25-36.
[9] Léger, P. and Leclerc, M. (2007). “Hydrostatic, temperature, time-displacement model for
concrete dams” J. Eng. Mech., 133(3), 267-277.
[10] Bofang, Z. (2003). “RCC arch dams: Temperature control and design of joints.” Int. Water
Power Dam Constr., 55(80, 26–30.
[11] Engineering monograph, (1977). “Design Criteria for concrete arch and gravity dams” Vol. 19.
[12] Yoshida, H., Yamazaki, M., and Shibata, K., (1983). “An analytical study on the thermal stress
of mass concrete” Part iv, The annual report of A IJ 110-117
[13] Janna, W., (1988). “Engineering heat Transfer”, SI Ed., Van Nostrand Reinhold
(international), UK
[14] Sabbagh-Yazdi, S.R. and Mastorakis, N.E. (2007). “3D heat generation and transfer in gravity
dam on rock foundation using Galerkin Finite Volume solver on tetrahedral mesh” International
Journal of Geology, 1(3), 52-60.
[15] U.S. Army Corps of Engineers (USACE). (1995).“Engineering and design; gravity dams.” EM
1110-2-2200, Washington, D.C.
[16] Tarbox, G. S. (1977). “Design of concrete dams.” Handbook of dam engineering, A. R. Golzé,
ed.,Van Nostrand Reinhold Company, Toronto, Canada.
[17] Ishiguro, S. and Nakaya, M. (1986). “Temperature distribution analysis of concrete in fill dam
gallery” Bull. Univ. Osaka, vol. 38, 35-43.
[18] Ashna, N. and Koichi, F. (2001). “Three dimensional finite element analysis of a Roller
Compacted Concrete (RCC) dam due to variable thermal loads” J. Civ. Eng. Bangladesh, 9(1), 65-
85