Advances in Engineering Software 38 (2007) 886895

www.elsevier.com/locate/advengsoft

Development of nite element computer code for thermal analysis

of roller compacted concrete dams
M.S. Jaafar a, K.H. Bayagoob a, J. Noorzaei a,*
, Waleed A.M. Thanoon b

a
Civil Engineering Department, University Putra Malaysia, 43400 Serdang, Malaysia
b
Civil Engineering Department, University Technology Petronas, 31750 Tronoh, Perak, Malaysia

Received 7 October 2004; accepted 14 August 2006

Available online 23 January 2007

Abstract

Thermal analysis of roller compacted concrete (RCC) dams plays an important role in their design and construction. This paper deals
with the development of a nite element based computer code for the determination of temperatures within the dam body. The nite
element code is then applied to the real full-scale problem to determine the impact of the placement schedule on the thermal response
of roller compacted concrete dam. Based on the results obtained, it could be concluded that for a given roller compacted concrete dam,
changing the placing schedule can optimize the locations of maximum temperature zones.
 2007 Elsevier Ltd. and Civil-Comp Ltd. All rights reserved.

Keywords: RCC dams; Birth; Death; Thermal analysis; Field problems; Schedule of construction; Finite element

1. Introduction to the design engineer. These include: limiting the placing

of the roller compacted concrete to the time of year when
The temperature in concrete arising from the hydration cool weather is expected, lowering the placing temperature,
of cement in roller compacted concrete (RCC) dams, placing at night, and jointing.
coupled with the low conductivity of concrete can induce To execute the thermal analysis, the nite elements
high thermal gradients in the interior and on the exterior should be added according to the casting schedule of the
surfaces of the dam. The internal restraints, due to this roller compacted concrete. This condition along with
thermal gradient and the external restraints, such as the above important parameters is taken into account by
foundation restraint, during the cooling of the dam can the developed nite element code in the present investiga-
cause signicant thermal stresses, which are sucient for tion. The birth and death of elements technique is used
cracking. So the analysis and design of roller compacted for simulating the dam construction schedule. It is a useful
concrete dams necessitate some forms of the thermal anal- technique for decreasing the volume of input data. Appli-
ysis to be performed [1]. The temperature distribution cability of the developed code is demonstrated by analyzing
through the dam and its evaluation with time, depends a real roller compacted concrete dam of 169 m in height.
on some thermal analysis parameters such as concrete
properties, ambient conditions, roller compacted concrete
placing temperature, lift thickness and in particular the
casting schedule of concrete. ACI Committee 207 [2], sug- 2. Mathematical model
gested a variety of options to minimize the thermal stresses
The Fourier equation governed the thermal generation
and temperature distribution is expressed by the following
*
Corresponding author. formula (for two dimensional and isotropic solid
E-mail address: [email protected] (J. Noorzaei). environments):

0965-9978/$ - see front matter  2007 Elsevier Ltd. and Civil-Comp Ltd. All rights reserved.
doi:10.1016/j.advengsoft.2006.08.040
 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895 887
   
o oT o oT oT evaluation of temperatures at the new time stage using
kx ky Q_ qc 1
ox ox oy oy ot the following expression;
where T is the concrete temperature, kx and ky are the con- 2
fT gb fT ga fDT g 8
crete conductivity coecients in x- and y-directions, 3
respectively, Q_ is the rate of heat of hydration introduced
per volume, q is the material density, and c is the concrete Since conductivity coecient (kx, ky), density (q), specic
specic heat. heat (c) are constants, the value of the [C] and [K] in the
Two main types of boundary conditions are Drichlet above expressions are also constant over each element.
and Cauchy boundary, which can be written, respectively, Hence the matrix [AG ] for each element has a constant
as [3,4]: value for each time interval. Therefore, the nodal tempera-
ture changes are only based on the value of the heat load
T Tp
oT oT 2
kx lx k y ly q hT s  T f Start
ox oy
where Tp is the known values of the nodal points of the Birth and Death of
temperatures on the boundaries; q is owing heat from sur- Elements Technique
face; h is the lm coecient; Ts is unknown temperatures at
the boundary nodal points; Tf is the ambient temperature; Read the Geometric, Material Properties
lx and ly are the direction cosines of the normal to the sur- and Boundary Conditions data
face under consideration [3,4].
Read the initial nodal temperatures
3. Numerical model

The numerical solution scheme used in this study is Generate the Thermal Matrices
based on the TaylorGalerkin approach. Upon applying [Kt]e, [C]e, {Ft}e, & [ AG* ]
this approach the following system of dierential equation
is obtained: - Compute the daily temperatures Tf
  - Compute the heat of hydration Q .
oT
Ce K t e fT g  fF t ge 0 3 - Compute the temperature values {T}b
ot
at the end of time step
where [C]e is the capacitance matrix; K t e is the heat sti-
ness (conduction and convection) matrix; and fF t ge is the Save the temperatures {T}b to use as the initial
total load heat vector due to hydration and convection temperatures {T}a for next time increment ti+1.
actions. Save the temperatures for selected nodes and stages
Save reference temperatures for structural analysis
The nite dierence approximation was used for solving
Eq. (3) in the time domain numerically. This solution (after
assembly of the stiness matrices) is given by [3] Evaluate the load heat vector matrix {FG}*

C hDtK t fT gb C  1  hDtK t fT ga

Assemble the element stiffness matrices
Dt1  hfF t ga hfF t gb 4 and load vectors
Solve the equation system to find the
where {T}b and {Ft}b are {T} and {Ft} at time (b) and fT ga temperature changes vector {T}
and fF t ga are {T} and {Ft} at time (a), h is a scalar
0 6 h 6 1 which is equal to 2/3 in Galerkin method. Then
Eq. (4) takes the following general form:
No End of stage
AG fDT g fF G g 5 Take a new time step placement

where Yes
No
1 2 Last Stage
AG  C K t  6 Take a new stage
Dt 3 Yes
and Save the temperatures for
selected nodes and stages
1
fF G g fF t ga 2fF t gb  3K t fT ga 7
2
End
where {DT} represents the temperature changes at the
nodal points with respect to time Dt which is used for Fig. 1. Program ow chart.
888 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895

vector fF G g. In other words the temperatures variations 4.2. Placing temperature

with time will depend only on the values of concrete
hydration rate (Q)_ and the ambient temperature (Tf). The temperatures of the aggregates used have great
The computational steps are summarized as follows: inuence on the initial temperature of fresh roller com-
pacted concrete. Since the average annual temperature is
Step i. Take the rst stage and divide the construction generally used instead of the aggregate temperature, the
time into several time increments (optimum time placing temperature is calculated as
step in terms of hours). 2
Step ii. Solve Eq. (5) and evaluate the changes of temper- T casting T anu  T anu  T mon T crush T mix T trans
3
atures {DT}.
9
Step iii. Accumulate the changes of temperature vector
{DT} with the initial concrete temperature vector where Tcasting is the roller compacted concrete placing tem-
using Eq. (8). perature, Tanu is mean annual air temperature at the dam
Step iv. The temperature vector of the ith time increment site, Tmon is the average monthly air temperature, Tcrush
{T}i will be the initial temperature vector for the is the added temperature of the crushed aggregate, Tmix is
next time increment Dti1 . the added temperature due to mixing, and Ttrans is the
Step v. Repeat steps iiv until the construction of the rst added temperature due to transportation.
stage is completed. The added temperatures due to aggregate crushing and
Step vi. Take the next stage in addition to the previous concrete mixing are assumed as 1.1 C in this investigation
stage and repeat steps iivi until the construction [7].
of the nal stage is completed.
5. Hydration heat
Based on the above computational procedure a 2-D
nite element code is written in Fortran language compat- Chemical reaction which occurs during the hardening of
ible with Fortran 90 power station. The ow chart of this the fresh concrete, i.e., hydration of cement, is a calorica-
program is as shown in Fig. 1. tion reaction and will cause some rising in the concrete vol-
ume. There are two models available in the literature for
4. Initial conditions simulation of the heat of hydration in concrete structures,
these are as follows.
The temperature distributions in the rock ground as well
as the roller compacted concrete placing temperature are
5.1. Adiabatic hydration model
the two initial conditions that need to be considered in
the analysis.
In most thermal stress modelling, an adiabatic hydration
model in which the heat is not lost to/or gained from the
4.1. Evaluation of foundation temperature
surrounding is used [5,9]. Typical adiabatic temperature
generation curves are given by ACI 207 [10] for dierent
Before calculating the temperature within the dam, the
placing temperatures (low to high-heat cements). Particular
temperature distribution in the rock ground, just before
adiabatic temperature rises and other thermal properties of
casting of the roller compacted concrete need to be evalu-
some roller compacted concrete mixtures are also given by
ated. As a method to calculate it, it is rst assumed that
ACI 207 [2]. The Japanese code of practice JCI 1986 [11]
the initial temperatures of all the nodes of the rock ground
gives the adiabatic curves parameters for a wide range of
are all same and equal to the main annual air temperature.
cement types placed at dierent temperatures.
Then the heat transfer is analyzed between the atmospheric
The adiabatic temperature rise of concrete is given by
temperature and the rock ground for two or three years
[5,11];
[5,6]. In this way, the temperature distribution of the rock
ground can be obtained. T ad K t 1  eat 10

Fig. 2. Birth and death of elements technique according to the casting schedule of RCC: (a) mesh generation, (b) activating bed rock, (c) birth of 1st layer,
(d) birth of other layers.
 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895 889

where Kt is the maximum temperature rise of concrete Conduction coefficient k = 401 W/m C
under an adiabatic condition; a is a parameter which repre- Specific heat coefficient c = 385 J/kg C
Material density = 8933 kg/m3 Perfect insulation
sents the heat generation rate and t is the time (h).
5 2
The cumulative heat generated due hydration up to time q = 310 W/m
100 mm
t is given by T0 T150

Q cqT ad 11 8 @ 75 mm

Using Eqs. (10) and (11), the expression for the rate of heat
of hydration to be used can be written as Fig. 3. Finite element mesh of the cooper thick wall.

Q_ cqK t aeat 12
FE (0 mm) Analytical (0mm)
The values of Kt and a are taken equal to 17 C and 0.0183, FE (150 mm) Analytical (150 mm)
respectively, in this investigation. 140
T0
Heat of hydration adiabatic calorimeter tests or large 120
concrete blocks with insulated surfaces can alternatively 100

Temp. (C)
be used when an accurate evaluation of the parameters is
80
required [9]. T150
60
Adiabatic hydration models give the fastest rate of heat
generation as they are based on the implicit criterion that 40
the heat already produced by hydration accelerates this 20
hydration process in turn. Although this may be valid to 0
the interior of massive structures such as dams, it would 0 20 40 60 80 100 120 140
become invalid near the surfaces of massive structures or Time (s)
in other members such as walls and foundations [9]. Fig. 4. FE and analytical solutions comparisons.

5.2. Non-adiabatic hydration model

Wp
Wb 14
Hydration of cement is a thermally-activated process e  ec
and consequently the rate of cement hydration depends
on the reaction temperature (i.e., at which the reaction where ec is the tensile strain capacity of the roller com-
takes place). This known as non-adiabatic hydration pro- pacted concrete.
cess in which the reaction temperature in turn depends Substituting for the value of e from Eq. (13), the allow-
on the heat-transfer characteristics within the concrete able temperature drop can be expressed as
mass. Therefore, the heat-of-hydration and the heat trans-  
1 Wp
fer processes are coupled. DT ec 15
However, investigators [2,5,10] suggested the use of adi- K RK f a W b
abatic model to simulate the heat of hydration in massive
concrete structures. Hence in this study the adiabatic
hydration model has been adopted.

6. Assessment of allowable temperature

In order to assess the allowable temperature gradient of

the dam cross-section which depends on the spacing of the
contraction joints, the coecient of thermal expansion, the
tensile strain capacity of the roller compacted concrete and
the eective restraint provided by the supporting media, a
simplied method is used herein in which the mass gradient
strain is calculated by the following equation:
e K R K f aDT 13
where e is the induced strain; KR, Kf, the structure and re-
straint factor, respectively; a, the coecient of thermal
expansion and DT is the temperature dierence with re-
spect to interior temperature dierence.
For an assumed permissible crack width (Wp), the dam
block width (Wb) can be calculated as Fig. 5. Dam cross-section.
890 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895

Fig. 6. Finite element mesh of RCC dam.

165
150
135
120
105
Height

90
75
60
45
30
15
0
0 2 4 6 8 10 12 14 16 18 20 22 22.5
Fig. 8. Boundary conditions.
Time (month)

Fig. 7. Height-time changes of RCC dam.

28

24
Eq. (15) is used in this investigation for assessing the allow-
able temperature drop of the roller compacted concrete. 20
Temp

16
7. Birth and death of elements technique
12
The incremental construction of roller compacted con-
crete (RCC) dam is modelled using the birth and death 8
0 20 40 60 80 100 120 140
algorithm. The application of this algorithm is shown in Time (day)
Fig. 2. Initially, the nite element mesh of the dam and Dt = 2 hr Dt = 3 hr Dt = 6 hr Dt = 12 hr
its foundation is generated as shown in Fig. 2a. Then the
rock ground elements are made active and are considered Fig. 9. Comparison of several time-intervals for the selected node (387).
for the thermal analysis (Fig. 2b). Next, the rst layer of
the dam body is made active and added to previous stage at one surface such that the net heat ux is maintained at
as shown in Fig. 2c. In the same way, the sequence of con- constant value of 3 05 W/m2. The example is solved by
struction of the dam is modelled using the above technique the developed nite element code using 8-noded isopara-
(Fig. 2d). metric element where a 600 mm of the slab thickness is
taken which is greater than the required thickness for the
8. Model verication validity of the assumed semi-innite media in the case of
2 minute exposure time. The idealization and the material
The code is veried by solving a few examples of heat properties used for solving this example are as shown in
transfer problems available in literature and only result Fig. 3. The solution obtained form the developed FE code
of one of these examples (given in Ref. [12]) is presented after an elapsed time of 2 min is compared with analytical
herein in which a thick slab of copper initially held at a uni- solution given in Ref. [12] for a two points located at the
form temperature of 20 C is suddenly exposed to radiation surface (T0) and 150 mm from the surface (T150). This
 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895 891

Table 1
on the U/S 5 m from U/S
Properties of the roller compacted concrete (RCC) material 39.5 m from U/S 74.5 m from U/S
45
Material properties Materials
40
Rock ground CMC RCC 35

Temp. (C)
Heat conduction coe. (k) kcal/m h C 1.3 1.81 1.81 30
Heat convection coe. (h) kcal/m2 h C 10.0 10.0 10.0 25
Specic heat (c) kcal/kg C 0.22 0.22 0.22 20
Material density (q) kg/m3 2710 2400 2400 15
10
5
0
30

15

30

45

60

75

90

10

12

13

15

16
00

00

00

00

00

00

50

50

00

50
00
0

0
0
0
25
Time (hr)
20
Temp

On the U/S 5 m form U/S

15
39.5 m from U/S 74.5 m form U/S
10
45
40
5 35

Temp. (C)
30
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 25
layer (each 20 days) 20
Mean Ambient Temp RCC Placing Temp 15
10
Fig. 10. RCC placing temperature for 1 November start. 5
0
0

15

30

45

60

75

90

10

12

13

15

16
00

00

00

00

00

00

50

00

50

00

50
0

0
0

0
30 Time (hr)
25
Fig. 12. Temperature histories at elevation 2.5 m under construction of
20 stage No. 1: (a) November 1 start schedule, (b) July 1 start schedule.

15
Temp

10
On the U/S 3.5 m from U/S
22.5 m form U/S 44.0 from U/S
5 45
0
40
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
layer (each 20 days)
Temp. (C)

30
25
Mean ambient Temp. RCC Placing Temp.
20
15
Fig. 11. RCC placing temperature for 1 July start. 10
5
0
solution is shown in Fig. 4 which is clear that it is in good
60

75

90

10

12

13

15

16
00

00

50
00

50
00

50

00

agreement with the analytical solution.

0
0

0
0

Time (hr)

On the U/S 3.5 m from U/S

9. Model application
22.5 from U/S 44.0 form U/S
45
From the available literature [13], the roller compacted 40
concrete dam shown in Fig. 5 is taken as a practical exam- 35
Temp. (C)

30
ple in this investigation. The maximum dam height is
25
169 m. The upstream and downstream facings are made 20
of conventional concrete. The cohesion layer on the bot- 15
tom is also made of conventional concrete (3 m thick). 10
Although the real roller compacted concrete is placed in 5
2530 cm thick horizontal layers, 5 m thick layers with the 0
60

75

90

10

12

13

15

16

placement time of 20 days are used in order to decrease the

00

50

00
00

50

50
00

00

0
0

computation time [8]. Hence the section of the roller com- Time (hr)

pacted concrete dam is divided into 34 layers given a total Fig. 13. Temperature histories at elevation 57.5 m under construction of
of 329 elements and 1029 nodes with one degree of freedom stage No. 12: (a) November 1 start schedule, (b) July 1 start schedule.
892 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895

32 Casting Temperature

28 After 24 hr

24 After 5 days

After 10 days
20
Temp. C

After 30 days (Ref. T.)

16
After 50 days
12
After 100 days
8
After 1 year

4 After dam construction

(680 days)
0
315 316 317 318 322
Node No. (from U/S to D/S)

46 Casting Initial Temperature

42 After 24 hr

38 After 5 days

After 10 days
34
Temp. C

After 30 days (Ref. T.)

30
After 50 days
26
After 100 days
22
After 1 year

18 After dam construction (680

days)
14
315 316 317 318 322
Node No. (from U/S to D/S)

Fig. 14. Temperature changes at elevation 2.5 m for the mid nodes of stage No. 1: (a) 1 November start, (b) 1 July start.

per node (Fig. 6). The progress of the dam construction mal analyses. The eect of solar radiation during the con-
with respect to time is illustrated in Fig. 7 while the bound- struction was incorporated by allowing an increase in
ary conditions throughout the dam construction are shown ambient temperature of 1.0 C to account for solar radia-
in Fig. 8. tion heating of the concrete surface [7].
The surface heat transfer coecient h (lm coecients)
9.1. Time step optimization in thermal equation is applied to all exposed surfaces to represent the convec-
tion heat transfer eect between the surrounding air and
Since the interval of the time step (Dt) in the thermal Eq. concrete surface. The following approximate equation is
(6) has a signicant eect on the accuracy of the results and used to calculate the surface heat transfer coecient [7];
computational eorts, number of trials have been carried
out to nd out the most optimum time step. h hc hw 16
Fig. 9 show the temperature variations at node 387 in
where for concrete surface the average value of hc is taken
the dam as indicated in Fig. 6 for dierent time interval
as hc = 4.84 kcal/m2 h C, and hw is related to the wind
(Dt). It is clear from this gure that the best interval of time
speed v approximately as hw = 1.086v, (v in km/h). In this
step is 3.0 h because it yields the same response as Dt = 2 h,
work, the monthly average wind velocities at the project
so there is no point to use a time step shorter than 3 h.
site are assumed uniform over the construction period
(680 days) and taken as v = 4.75 km/h. Thus using Eq.
9.2. Material properties
(16), yields h = 10.0 kcal/m2 h C.
The material properties for the conventional mass con-
crete (CMC) and the roller compacted concrete (RCC) 9.3. Placing schedule
adopted for nite element analyses are summarized in
Table 1 which are taken from Saetta et al. [8]. The eect of dierent placement schedules on the ther-
Based on the observed data at the project site, the daily mal response of the roller compacted concrete dam has
changes of temperatures are taken into account in the ther- been investigated. The rst schedule starts on the 1st
 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895 893

40
Casting Initial Temperature

After 24 hr
36
After 5 days
Temp C
32 After 10 days

After 30 days (Ref. T)

28
After 50 days

24 After 1 year

After dam construction (680 days)

20
568 569 570 571 575
Node No. from U/S to D/S

32
Casting Initial Temperature
28
After 24 hr
24
After 5 days
20
Temp C

After 10 days
16
After 30 days (Ref. T)
12
After 50 days
8
After 1 year
4
After dam construction (680 days)
0
568 569 570 571 575
Node No. from U/S to D/S

Fig. 15. Temperature changes at elevation 57.5 m for the mid nodes of stage No. 12: (a) 1 November start, (b) 1 July start.

November while the second schedule begins on the 1st July the 1st stage has been observed. However, the temperature
as shown in Figs. 10 and 11, respectively. remains almost constant at the interior nodes (22.5 and
44.0 m from U/S). The maximum temperature predicted
9.4. Temperatures distribution for the 1st Nov. start and the 1stJuly start are 42 C and
30 C, respectively. This clearly indicates the eect of con-
Fig. 12a and b shows the temperature histories for few struction schedule on the temperature variations along the
nodes which are located at elevation of 2.5 m correspond- dam height which followed by the seasonal changes.
ing to Stage 1. These nodes are situated at distances of Fig. 14a and b shows the temperature changes for some
0.0, 5.0, 39.5 and 74.5 m from the upstream face, respec- nodal points of stage 1 for the two starting schedules (1
tively, as shown in Fig. 6 (Nodes: 315, 316, 317, 318). Ini- November start and 1 July start) after several time periods
tially, just after concreting starts, there is a sudden rise in of the roller compacted concrete placement. The positions
temperatures at all nodes. Then the temperatures at the of all nodal points are shown in Fig. 6. Initially, all the
nodes on the dam surface follows the ambient temperature nodes posses the same temperature degree (placement tem-
(sinusoidal shape), while there is a drop in temperature at perature). Later, there is an increase in the temperature of
the interior nodes takes place with a slow rate. The maxi- interior nodes during the rst 10 days due to the eect of
mum temperature predicted for 1 Nov. start and 1 July the heat of hydration. After 10 days, the heat of hydration
start are 32 C and 42 C, respectively. stopped and the temperature at these nodes starts to
The temperature histories for four nodal points which decreased gradually. Whereas for those nodes which are
are located at elevation of 57.5 m while Stage 12 is under located on the upstream and downstream, the temperature
construction are shown in Fig. 13a and b. These nodes changes follow the ambient temperature changes.
are situated at distances of 0.0, 3.5, 22.5 and 44.0 m from Fig. 15a and b shows the temperature changes for the
the upstream face, respectively, as shown in Fig. 6 (Nodes: mid nodes of stage 12 for the two starting schedules (1
568, 569, 570, 571). Similar trend of behaviour to that of November start and 1 July start) after several time periods
894 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895

Fig. 16. The isothermal curves of the RCC dam at the end of stage No. 20.

Fig. 17. The isothermal curves at the completion of the RCC dam construction.

of the roller compacted concrete placement. Similar trend (iii) The value of the tensile strain capacity (ec) is assumed
of behaviour to that of 1st stage for the temperature equal to 80 lm.
changes with time has been observed. (iv) The permissible crack width is assumed equal to of
Furthermore an attempt has been made to illustrate the 2.0 mm.
variation of temperature in the dam body for both starting
time schedules. Figs. 16 and 17 shows the contour plots of For the present case of study, the dam block width (Wb)
the temperature distribution in stage 20 and 34 under, being analyzed is taken as 20 m, substituting the adopted
respectively. It is clear from those plots that there is a little values of the parameters (Wb), (Wp), and ec in Eq. (15)
dierence in the maximum temperature values, but the yields an allowable temperature drop of 18.0 C. From
location of these maximum values is at dierent zone. the safety point of view, the comparison between the allow-
able temperatures drop for the two schedules at the dam
9.5. Allowable temperature drop base, for the 1st July starting schedule, the maximum tem-
perature rise is 42 C and the roller compacted concrete
To determine the allowable temperature drop, the fol- placing temperature is 25 C yields a maximum temperature
lowing parameters are adopted: drop of 17 C while for the 1st November starting schedule,
the maximum temperature rise is 28 C and the roller
(i) The restraints coecients KR and Kf are conserva- compacted concrete placing temperature is 15 C yields a
tively assumed equal to 1.0 [1]. maximum temperature drop of 13 C. Both schedules
(ii) The value of the coecient of thermal expansion a is yield a temperatures drop values lesser than the allowable
taken as 1 105. drop temperature. The November starting schedule drop
 M.S. Jaafar et al. / Advances in Engineering Software 38 (2007) 886895 895

temperature which is 13 C is less than that of the 1st July compacted concrete dams carried by a Ph.D. candidate
starting schedule (17 C). Thus, the November schedule is K.H. Bayagoob in the Civil Engineering Department, Uni-
much safer than the July schedule in which the probability versity Putra Malaysia.
of cracking are more likely to occur due to its drop temper-
ature is very close to the allowable drop temperature.
References
10. Conclusion [1] Tatro S, Schrader E. Thermal analysis for RCC a practical
approach. Roller compacted concrete III. New York: ASCE; 1992,
Based on the above study the following conclusions can p. 389406.
be drawn: [2] ACI 207 5R-99. Roller-compacted mass concrete. ACI manual of
concrete practice. Part 1, USA; 2004.
[3] Sergerlind LJ. Applied nite element analysis. New York: John
(i) By using the birth and death of elements technique, Wiley and Sons; 1984.
investigation of dierent casting schedules can be car- [4] Hinton E, Owen DRJ. An introduction to nite element computa-
ried out. tions. UK: Swansea: Pineridge Press Limited; 1981.
(ii) The starting date of the roller compacted concrete [5] Ishikawa M. Thermal stress analysis of a concrete dam. Comput
Struct 1991;40(2):34752.
placement schedule has signicant eect on the distri-
[6] Araujo JM, Awruch AM. Cracking safety evaluation on gravity
bution of temperatures in the roller compacted con- concrete dams during the construction phase. Comput Struct
crete dam body. Starting the roller compacted 1998;66(1):93104.
concrete placement in hot season leads to the devel- [7] US Bureau of Reclamation. Level 2 thermal study mass gradient and
opment of higher temperature zones in the lower por- surface gradient analysis procedure and examples, ETL 1110-2-542;
1997.
tion of the dam body.
[8] Saetta A, Scotta R, Vitaliani R. Stress analysis of concrete structures
(iii) Changing the placing schedule can optimize the loca- subjected to variable thermal loads. J Struct Eng ASCE 1995;
tion of maximum temperature zones which might be 121(3):44657.
leads nally to decreasing the tensile stresses at the [9] Vitharana N, Wark R. Thermal cracking in large concrete place-
critical zone. ments: theory and applications. In: ANCOLD conference; 2002. p. 1
12.
(iv) The allowable temperature drops obtained in this
[10] ACI 207 1R. Mass concrete for dams and other massive structures.
investigation indicates that, the November starting ACI Committee, USA; 1987.
schedule is safer than the July starting schedule. [11] Japan Concrete Institute. Standard specications for design and
construction of concrete structures. Part 2 (construction), Tokyo;
1986.
[12] Incropera Frank P, DeWitt David P. Fundamentals of heat and mass
Acknowledgement
transfer. New York: John Wiley & Sons; 2002.
[13] Noorzaei J, Meherdadi M. Thermal analysis of roller compacted
This investigation is a part of the in progress Ph.D. work concrete dams. In: World conference on concrete materials and
under the title of Thermal and structural analysis of roller structures, Malaysia; 2002. p. 38194.