Content uploaded by Serife Yurdagul Kumcu
Author content
All content in this area was uploaded by Serife Yurdagul Kumcu on Jan 09, 2018
Content may be subject to copyright.
KSCE Journal of Civil Engineering (0000) 00(0):1-10
Copyright ⓒ2016 Korean Society of Civil Engineers
DOI 10.1007/s12205-016-1257-z
−1−
pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
Water Engineering
Investigation of Flow Over Spillway Modeling and Comparison
between Experimental Data and CFD Analysis
Serife Yurdagul Kumcu*
Received May 14, 2014/Revised 1st: August 13, 2015, 2nd: January 26, 2016/Accepted April 24, 2016/Published Online June 24, 2016
··································································································································································································································
Abstract
As a part of design process for hydro-electric generating stations, hydraulic engineers typically conduct some form of model
testing. The desired outcome from the testing can vary considerably depending on the specific situation, but often characteristics such
as velocity patterns, discharge rating curves, water surface profiles, and pressures at various locations are measured. Due to recent
advances in computational power and numerical techniques, it is now also possible to obtain much of this information through
numerical modeling. In this paper, hydraulic characteristics of Kavsak Dam and Hydroelectric Power Plant (HEPP), which are under
construction and built for producing energy in Turkey, were investigated experimentally by physical model studies. The 1/50-scaled
physical model was used in conducting experiments. Flow depth, discharge and pressure data were recorded for different flow
conditions. Serious modification was made on the original project with the experimental study. In order to evaluate the capability of
the computational fluid dynamics on modeling spillway flow a comparative study was made by using results obtained from physical
modeling and Computational Fluid Dynamics (CFD) simulation. A commercially available CFD program, which solves the
Reynolds-averaged Navier-Stokes (RANS) equations, was used to model the numerical model setup by defining cells where the flow
is partially or completely restricted in the computational space. Discharge rating curves, velocity patterns and pressures were used to
compare the results of the physical model and the numerical model. It was shown that there is reasonably good agreement between
the physical and numerical models in flow characteristics.
Keywords: dam structures, spilllway modeling, CFD anaysis, numerical model, physical model
··································································································································································································································
1. Introduction
Hydraulic design of a spillway and a stilling basin has been
one of the most studied subjects in hydraulic engineering.
Properly designed approach flow conditions, spillways and
stilling basins will be able to pass flood flows efficiently and
safely to downstream of dams. Physical scale modeling has been
used in the design and investigation of hydraulic structures for
over 100 years. A hydraulic model is still a precision device for
the experimental investigation of flow over a spillway structure,
which can give reliable information only if it is designed correctly
(Willey et al., 2012).
With increasing computer processing capacity, numerical
simulations for hydrodynamic processes become attractive,
including flow over spillways. A comparison of these numerical
results with experimental or prototype data is still required for
calibration and validation. Computational Fluid Dynamics (CFD) is
a branch of numerical modeling that has been developed for
solving problems involving fluid flow. This includes applications
involving fluid-solid interaction, such as the flow of water in a
river or over and around hydraulics structures. There is therefore
considerable interest on the part of hydraulic engineers into the
applicability of CFD to model fluid flow. The majority of available
literature on applying CFD to spillway modeling came from studies
using Flow-3D, which solves the Reynolds-averaged Navier-
Stokes (RANS) equations (Ho et al., 2003; Savage et al., 2001;
Kim, 2010; Chanel, 2008). To track the free water surface (air/
water interface) as it moves in time and space, the program
implements a sophisticated algorithm called the Volume-of-Fluid
(VOF) method. The VOF method tracks the amount of fluid in
each computational cell. Cells range from completely empty to
completely full and the volume changes as flow moves in or out
from neighboring cells. Flow-3D also uses a Fractional Area/
Volume Obstacle Representation (FAVOR) method to define
obstacles (Hirt and Sicilian, 1985). This method allows Flow-3D to
use fully structured computational grids that are much easier to
generate than the deformed grids used by most other CFD
programs. Teklemariam et al. (2001) prepared a report outlining the
use of Flow-3D and it discusses how Flow-3D was successful in
matching the physical model test results for discharge as well as
flow patterns and velocities in modeling (Flow-3D User’s manual,
2012). The report also introduced the ability of Flow-3D to provide
discharge measurements that were very close to empirical
estimates. Teklemariam et al. (2008) discuss the great potential for
TECHNICAL NOTE
*Assistant Professor, Civil Engineering Dept., Necmettin Erbakan University, Konya, Turkey (Corresponding Author, E-mail: yurdagulkumcu@gmail.com)
Serife Yurdagul Kumcu
−2−KSCE Journal of Civil Engineering
the use of CFD for the assessment of a design, as well as screening
and optimizing of hydraulic structures and cofferdam layouts. They
conclude that CFD has been successful in optimizing the final
conceptual configuration for the hydraulics design of the project,
but recommend that physical modeling still be used as a final
confirmation.
This paper provides experimental studies performed on Kav ak
Dam and analyses the stability of spillway design by using
FLOW-3D model. It compares the hydraulic model tests with
FLOW-3D simulation results and gives information on how
accurately a commercially available Computational Fluid Dynamic
(CFD) model can predict the spillway discharge capacity and
pressure distribution along the spillway bottom surface.
2. Physical Model
A 1/50-scaled undistorted physical model of the Kavsak Dam
spillway and stilling basin was built and tested at the Hydraulic
Model Laboratory of State Hydraulic Works of Turkey (DSI).
The model was constructed of plexiglas and was fabricated to
conform to the distinctive shape of an ogee crest. The spillway
has 45.8 m in width and 57 m long with a bottom slope of 125%.
The length of the stilling basin is about 90 m. During model tests,
flow velocities were measured with an ultrasonic flow meter.
Pressures on the spillway were measured using a piezometers
s
ç
Table 1. Upstream and Downstream Operating Conditions of the
Kavsak Dam
Run Upstream reservoir
elevation (m)
Downstream tailwater
elevation (m)
1 306.55 168.00
2 311.35 174.50
3 314.00 178.90
4 316.50 182.55
Fig. 1. (a) Original Project Design and Final Project Design after Experimental Investigations and Flow Measurement Sections at the
Approach, (b) Top View Experimentally Modified Approach in the Laboratory, (c) Side View of the Experimentally Modified Approach
in the Laboratory
Investigation of Flow Over Spillway Modeling and Comparison between Experimental Data and CFD Analysis
Vol. 00, No. 0 / 000 0000 −3−
board reading provided the average pressure reading at each
pressure tap location. Both the upstream reservoir lake level and
downstream tailwater elevations were measured using piezometers.
A control valve was used to set the flow in the physical model.
The model was operated at four different upstream reservoir
elevations as given in Table 1. The 3rd and 4th runs in Table 1
belong to corresponding Q1000 and Q10000 discharge values of the
project. The downstream tailwater elevations was adjusted by
another control gate located far downstream of the model
(Kumcu, 2010).
3. Approach Flow Conditions
In order to obtain uniform flow conditions and decreasing the
energy losses through the approach, velocity must be less than 5
m/s ( entürk, 1994). As the flow velocities are high and flow is
vortex flow, the streamlines are separated in front of the gate
piers and the flow is not uniform. In order to obtain uniform flow
conditions some modifications were made in the reservoir
topography. The piers were extended into the dam reservoir and
their shapes were changed to pass the flow safely. As the piers
behave as a console they are also extended longitudinally in
vertical direction up to mixing with the dam body (290 m above
the talweg) for stability issues. Flow measurement sections and
modifications which were done after experimental investigations
are given in Fig. 1. Flow velocity measurements belonging to
original project and experimentally modificated project are given
in Fig. 2. It is clear from the Fig. 2 that, approach flow velocities
are decreasing after doing modifications. In the theoretical study
of the project, the discharge capacity of the spillway was
computed considering that all gates would be completely open
for all reservoir water levels. Discharge rating curve varies withS
ç
Fig. 2. Flow Velocity Measurements at the Approach Belonging to Theoretical Project Design and the Design which is Observed after the
Experimental Modifications
Serife Yurdagul Kumcu
−4−KSCE Journal of Civil Engineering
the water level over the crest. As the spillway rating curve is an
important parameter to show the consistency and accuracy of the
model application to prototype, the spillway rating curves obtained
from the theoretical and experimental studies are compared with
each other. According to Savage and Johnson (2001) the theoretical
flow discharge through a spillway can be expressed by:
(1)
where Q denotes the flow discharge, C discharge coefficient, L
the spillway crest length or width and H upstream head measured
from the crest to the unaffected upstream water stage. The
discharge coefficient, C is not constant. It is influenced by a
variety of factors including the depth of approach, relation of the
actual crest shape to the ideal nappe shape, upstream face slope,
downstream apron interference, and downstream submergence
(Design of small dams, 1977). Each of the conditions has ranges
and design curves that describe the effect of the parameters
mentioned. The theoretical flow discharges were calculated for
various water depths to obtain the rating curve over the spillway
by using Eq. (2). Experimental measurements were also gained
by using gauging of water depths for a given discharge. Comparison
of theoretical discharge measurements with experimental study
is shown in Fig. 3. It was seen in Fig. 3 that, although both values
are very close to each other, when the discharge capacity is
higher than Q= 3000 m3/s. For a given discharge, the flow can
pass over the spillway with lower flow depths than the
theoretical calculations, so it is in the safe side.
4. Pressure and Velocity Measurements on Spill-
way
Cavitation depends on the pressure and speed of flow, which is
shown by the Cavitation number. Office of the United States
claims an paper under the title “Cavitation on spillways and
chutes” (1990), presented the Cavitation Number as follows:
(2)
where; σ is the dimensionless cavitation number; P is the
absolute pressure (N/m2); Pu is the vapor pressure of the fluid (N/
m2); r is the flow density (kg/m3) and U is mean flow velocity
(m/s). If the flow pressure drops below the vapor pressure the
water begins to vaporize, just like if it was boiling (Wagner,
1967). As the bubbles can't escape they will implode. The
pressures fluctuations associated with bubble collapse produce
noise and vibrations and, eventually can lead to the structure
failure (Vischer and Hager, 1997). Cavitation can cause damage
in a very short time (Falvey, 1990). The cavitation number at
which cavitation starts is here referred as the critical cavitation
number (Kokpinar and Gogus, 2002). According to Erfanian-
Azmoudeh and Kamanbedast (1990) the critical cavitation
number is σ = 0.25, and cavitation will occur if σ< 0.25.
In order to investigate the flow characteristics and to calculate
the cavitation numbers; flow depths, the pressures and flow rates
were measured at Km: 22.5, 30.5 and 43.0 (Spillway crest is
Km: 0 + 000.00) for the reservoir discharge capacities of Q = 500
m3/s, 1000 m3/s, 2500 m3/s, 3856 m3/s and 5053 m3/s considering
that for spillway gates are fully open. Flow depths were measured
directly, and pressures were measured by using manometers. The
details of the experimental study and the calculations are given in
the report of Hydraulic Model Studies of Kavsak Dam and HPP
(Kumcu, 2010).
In the original design case, when the flow discharge is 3856
m3/s and 5053 m3/s, some negative pressures were attended
along the spillway and cavitation numbers are calculated. These
results indicate that the velocities were very high and the
pressure variations strongly affect the flow in this area. Flow
characteristics which were observed at these points presented in
Table 2. It is clearly seen that, cavitation numbers are under the
critical cavitation number, σ= 0.25 at Km: 0+043.00 for
Q = 3856 m3/s and 5053 m3/s. Therefore the cavitation risk is
real. Protection of the surface from cavitation erosion is usually
QCLH
3
2
---
=
σPPv–
1
2
---ρU2
---------------
=
Fig. 3. Rating Curves of Theoretical Project Design and the Design
which is Observed after the Experimental Modifications
Table 2. Flow Characteristics Along the Spillway from the Physical
Model
Cross-section
No
Distance from
Spillway crest
(Km)
Q
(m3/s)
U ave
(m/s)
Cavitation
Number
1 0+022.50 500 20.67 0.413
2 0+030.50 500 15.04 0.923
3 0+043.00 500 15.47 0.870
1 0+022.50 1000 18.21 0.544
2 0+030.50 1000 18.38 0.632
3 0+043.00 1000 20.51 0.509
1 0+022.50 2500 19.14 0.493
2 0+030.50 2500 20.40 0.549
3 0+043.00 2500 26.26 0.329
1 0+022.50 3856 20.30 0.429
2 0+030.50 3856 20.17 0.583
3 0+043.00 3856 27.78 0.218
1 0+022.50 5053 20.33 0.415
2 0+030.50 5053 21.72 0.517
3 0+043.00 5053 29.20 0.212
Investigation of Flow Over Spillway Modeling and Comparison between Experimental Data and CFD Analysis
Vol. 00, No. 0 / 000 0000 −5−
achieved by introducing air next to the flow structure surface.
Aeration devices are designed to introduce artificially air within
the flow upstream of the first location where cavitation damage
might occur. Aerators are designed to deflect high velocity flow
away from the chute surface. The water taking off from the
deflector behaves as a free jet with a large amount of interfacial
aeration. This is made by means of aeration devices located on
the bottom and sometimes on the sidewalls of the structure
(Chanson, 2002). If the gate piers are not going on longitudinally
along the spillway and the spillway is steep and short, aeration
slot located at the downstream end of the gate piers is used for
aeration (Chanson, 2005). In this design tip of the gate piers also
increases. In order to prevent cavitation, aerators were fixed at
the downstream end of the gate piers. In order to increase the
amount of air introduced by aerators the deflectors were placed
in front of the aerators (Demiröz, 1986). The plan view of
experimental arrangement of aerators and the damps is shown in
the Fig. 4.
It can be clearly seen from Fig. 5 that, aerators and deflector
are working very efficient, and face of the spillway has been
aerated enough by aerators.
5. Numerical Simulation
After serious modification are made on the original project, a
comparative study was made for flow over a spillway structure
using results obtained from 1/50 scaled physical modeling with
full scaled (prototyped) Computational Fluid Dynamics (CFD)
simulation. The commercially-available CFD package FLOW-
3D Version 10.0 was used in the simulation of the flow field. The
CFD package applies finite-volume method to solve the RANS
equations.
One of the main characteristics of turbulent flow is fluctuating
velocity fields. These fluctuations cause mixing of transported
quantities like momentum, energy and species concentration and
thereby also fluctuations in the transported quantities. Because of
the small scales and high frequencies of the fluctuations they are
too computationally expensive to simulate directly in practical
engineering situations. Instead, the instantaneous governing
equations are time-averaged to remove the small scales and the
result is a set of less expensive equations containing additional
Fig. 4. Plan View and Section View of the Aerator and Damp
Fig. 5. Aerators and Damps: (a) Shape of the Aerators and Deflectors (Chanson, 2005), (b) Application of the Aerators and Deflectors in
the Experimental Study
Serife Yurdagul Kumcu
−6−KSCE Journal of Civil Engineering
unknown variables. These unknown (turbulence) variables are
determined in terms of modeled variables in turbulence models.
This process of time-averaging is called Reynolds averaging.
When this is done the solution variables in the instantaneous
Navier-Stokes equations are decomposed into the mean (time-
averaged) and fluctuation components (Reynolds decomposition,
Margeirsson, 2007). For the velocity components:
(3)
where and are the mean and fluctuating velocity components
respectively. Scalar variables are decomposed in a similar way.
When expressions of this form for the flow variables are substituted
into the instantaneous continuity and momentum equations and a
time (ensemble) average is taken the Reynolds-averaged Navier-
Stokes (RANS) equations are yielded. They can be written as;
(4)
(5)
Here the overbar on the mean velocity has been dropped. The
velocities and other solution variables now represent time-
averaged values instead of instantaneous values. The additional
terms that have appeared are called Reynolds stresses
and must be modeled in order to close Eq. (5). Because
turbulence is the main cause of entrainment, a turbulence-
transport model must be used in connection with the air-
entrainment model and the traditional k-epsilon (ε) turbulence
model be employed in the present study. It is one of the two-
equation models which are considered the simplest of the so
called complete models of turbulence. Ever since it was proposed
in 1972 its popularity in industrial flow simulations has been
explained by its robustness, economy and reasonable accuracy
for a wide range of turbulent flows. The details of the k-ε model
can be found in Kim (2007). Free surfaces are modeled with the
Volume of Fluid (VOF) technique, which was first reported in
Nichols and Hirt (1975), and more completely in Hirt and
Nichols (1981). Trademarked as TruVOF, this technique is one of
the defining features of the program and provides three important
functions for free surface flow: location and orientation of free
surfaces within computational cells, tracking of free surface
motion through cells, and a boundary condition applied at the
free surface interface.
The location of the flow obstacles is evaluated by the program
implementing a cell porosity technique called the fractional area/
volume obstacle representation of FAVOR method (Hirt, 1992).
The free surface was computed using a modified volume-of-fluid
method (Hirt and Nichols, 1981). For each cell, the program
calculates average values for the flow parameters (pressures and
velocities) at discrete times using staggered grid technique
(Versteeg and Malalasekera, 1996). On the Cartesian coordinate
system (x, y, z), governing equations for analysis of incompressible
three dimensional flow are given below by Kim et al. (2010):
(6)
Where, are velocity components in the coordinate
directions , are fractional areas open to flow
in the coordinate directions of , ρ is density and RSOR is a
density source term.
(7a)
(7b)
(7c)
where, VF is a fractional volume open to flow, p is pressure,
are body accelaration in the coordinate direction
, and are viscous accelerations in the coordinate
direction .
The VOF method is based on the assumption that the two
fluids are not interpenetrating. Each phase (fluid) is given a
variable that accounts for how much percentage of each
computational cell is occupied by the phase. This variable is
called a volume fraction of the phase. The volume fractions of all
phases sum up to unity in each computational cell. The fields for
all variables and properties are shared by the phases and represent
volume-averaged values. Variables and properties represent either
only one phase or a mixture of phases in a given cell depending
on the volume fractions of the phases in the cell. If is
equal to 1, the control volume will be full of fluid, and F is equal
to 0, no fluid will exist in a control volume. Furthermore, in the
case of a free water surface, F is shown to have the value
between 0 and 1. Applying function F to Eq. (6) governing
equation becomes:
(8)
FLOW-3D version V10.0 was used to simulate flow over the
Kavsak Dam along with the renormalized group turbulence
model. A rectangular grid was defined in the computation
domain shown in Fig. 6. Total number of grid cells was
approximately 6.24E+06 in which only 4.36E+06 of them were
active. The corresponding uniform mesh size used in meshing
was ∆x = ∆y = ∆z = 0.5 m with rectangular grid system and 40-hr
elapsed computational duration. There were many tests have
been conducted to validate the usefulness of the air-entrainment
model. In each one, after finding the mesh size for each case, the
air entrainment rate coefficient was selected as Cai r = 0.5. Souders
and Hirt (2004) also indicated that, the value of air entrainment
coefficient, Cair, is expected to be less than unity because only a
portion of the raised disturbance volume is occupied by air. Cair =
0.5 assumes on average that air will be trapped over about half
uiuiu′
i
+≡
uiui
′
∂ρ
∂t
------∂
∂xi
------ ∂ui
()+0=
∂
∂t
---- ρui
()
∂
∂xj
------ ρuiuj
()+∂p
∂xi
------
–∂
∂xj
------ u∂ui
∂xj
-------∂uj
∂xi
-------2
3
---δij
∂ul
∂xl
-------–++=
+∂
∂xj
------ ρui′u′j
–()
ρui′u′j
–()
VF
∂ρ
∂t
------∂
∂x
----- uAx
()
∂
∂y
----- vAy
()
∂
∂z
----- wAz
()+++ RSOR
ρ
----------
=
uvw,,()
xyz,,()AxAxAx
,,()
xyz,,()
∂u
∂t
------1
VF
------uAx
∂u
∂x
------vAy
∂u
∂y
------wAz
∂u
∂z
------++
⎝⎠
⎛⎞
+1
ρ
---∂p
∂x
------– Gxfx
++=
∂v
∂t
----- 1
VF
------uAx
∂v
∂x
----- vAy
∂v
∂y
----- wAz
∂v
∂z
-----++
⎝⎠
⎛⎞
+1
ρ
---∂p
∂y
------– GyGyfy
+++=
∂w
∂t
-------1
VF
------uAx
∂w
∂x
-------vAy
∂w
∂y
-------wAz
∂w
∂z
-------++
⎝⎠
⎛⎞
+1
ρ
---∂p
∂z
------– Gzfz
++=
GxGyGz
,,()
xyz,,() fxfyfz
,,()
xyz,,()
Fxyzt,,,()
∂F
∂t
------ 1
VF
------∂
∂x
----- FuAx
()
∂
∂y
----- FvAy
()
∂
∂z
-----FwAz
()+++0=
Investigation of Flow Over Spillway Modeling and Comparison between Experimental Data and CFD Analysis
Vol. 00, No. 0 / 000 0000 −7−
the surface area. Usta E. (2014) selected air entrainment rate
coefficient 0.5 as the value which is suitable for most cases
recommended by Flow 3D User Manuel, 2012.
To simulate given flow, it is important that the boundary
conditions accurately represent what is physically occurring.
Because the flow is defined in Cartesian coordinates, there are
six different boundaries on the computational mesh domain. The
boundary conditions on the mesh were set as follows: sidewalls
y- common no slip, non-porous/wall; top z-pressure boundary
with gauge pressure equal to zero (atmospheric); bottom z-no
slip/wall; left x-local stagnation pressure based on upstream total
head H over the spillway crest with a hydrostatic pressure
distribution; and right x-local static pressure based on downstream
tailwater elevation with a hydrostatic pressure distribution. 50 m
upstream and 200 m downstream of the spillway crest were used
as the boundary conditions of left x and right x, respectively.
In running the FLOW-3D CFD software, computation modules
of viscosity and turbulence, gravity, air-entrainment, and density
evaluation were activated for all cases studied. Since there are no
prototype data available for comparison to the CFD solution, the
data from the physical model have been scaled to prototype
dimensions.
6. Discussion and Results
The main purpose of this part of the studies to compare results
from a physical model with that of a CFD model for flow over an
ogee crest spillway and through stilling basin. The flow rates
over the spillway crest and free surface elevations, depth-
averaged velocity distributions, and the pressures acting on the
crest and on the stilling basin are used to compare the differences
between the physical model and the CFD model. The existing
Kavsak Dam physical model data have been used as a baseline
of this comparison (Kumcu, 2010).
Table 2 shows the physical model measured flow rates (QPM)
and the numerically calculated flow rates from the CFD model
(QCFD). The results have been normalized to allow a comparison
in their simplest form in Fig. 4. The 10000 years return period
parameters, (H0)10000 = 16.46 m and Q10000 = 5053 m3/s, from
physical model are used as the basis. In Fig. 7 the static head
above crest, H0, is normalized by the (H0)10000 and shown in the
abscissa. The discharge Q is normalized by Q10000 and shown on
the ordinate. Using the physical model and its discharge as
observed standard, the relative percent difference in discharge is
calculated in Table 3. The relative percent difference at a given
(H0)/(H0)10000 is defined as (QCFD - QPM)/QPM x100 and shows that
the CFD model agrees within 3.2% in average with the physical
model.
The data for Q10000 = 5053 m3/s in the physical model was used
for the comparison of free surface elevation between the physical
model and the CFD model as seen in Fig. 8. Since similar results
Table 3. Comparison of Observed Flow Rate Versus Computed
Flow Rate (prototype scale)
Run QPM
(m3/s)
QCFD
(m3/s)
Percent
difference
1 1000 1034 3.4
2 2500 2415 3.4
3 3856 4001 3.7
4 5053 5170 2.3
Fig. 6. (a) Solid Model, (b) Solid Model with Mesh of the Kavsak
Dam Spillway used in the CFD Simulations (final design)
Fig. 7. Comparison between the Physical Model (PM) and the
Numerical Model (CFD) Predictions for Flow Rates Over
Spillway
Fig. 8. Comparison of Free Surface Elevations between Physical
Model and CFD Model for Q= 5053 m3/s
Serife Yurdagul Kumcu
−8−KSCE Journal of Civil Engineering
were obtained, other simulation plots and comparison with
physical model data will not be given here. In the comparison,
free surface data is plotted in elevation where the crest is at 300
m above the sea level. As seen in the figure, the majority of the
points overlap exceptionally well, while only the hydraulic jump
roller region on the profile seems to exhibit any notable error.
This is due to the difficulties of both CFD modeling and
measurement in physical model in accounting effects of strong
turbulence at the hydraulic jump region and flow aeration with
related consequences on bulking of flow depth.
Figures 9 shows 2D view of depth-averaged velocity contours
obtained from the CFD model for the flow rate of Q = 5053 m3/s.
Since most of the free surface elevation data of physical model
overlaps the CFD model data, the depth-averaged velocity
values of both models also show similarities. The maximum
value of depth-averaged velocity was found as approxiamtely 32
m/sec which creates a potential risk for cavitation damage.
The distiribution of air antrainment rate obtained from the
CFD model data along with the flow over spillway and through
stilling basin for the flow rate of Q = 5053 m3/s was given in
Fig. 9. CFD Solution of 2D Depth-averaged Velocity Distribution
Along the Spillway Structure for Q10000 = 5053 m3/s (velocity
values are in m/s)
Fig. 10. 2D-CFD Solution of Volume Fraction of Entrained Air Con-
tours Along the Spillway Structure for Q10000 = 5053 m3/s
Fig. 11. Comparison of CFD and Physical Model Pressures for: (a) Q = 1000 m3/s, (b) Q = 2500 m3/s, (c) Q = 3856 m3/s, (d) Q = 5053 m3/s
Investigation of Flow Over Spillway Modeling and Comparison between Experimental Data and CFD Analysis
Vol. 00, No. 0 / 000 0000 −9−
Figure 10. According to the Fig. 7, the bottom surface of the
spillway downstrean of the aerator structure, where potentially
under the risk of cavitation damage, is sufficiently aerated. Since
the value of 1-3% of air concentration can be generally accepted
as a critical value for the prevention of cavitation damage, CFD
results promise always more than 10% of air conenration value.
With the horizontal distance starting from the crest axis the
bottom pressure distributions (in Pascal) along the spillway and
stilling basin have been shown on Figs. 8 to provide a comparison
of spillway and stilling basin average pressures for four different
flow rate conditions on the physical model as; Q = 1000 m3/s,
2500 m3/s, 3856 m3/s, and 5053 m3/s. Pressures from the CFD
model compared quite favorably with the scaled physical model
data with the exception of pressure data obtained around baffle
blocks located at the longitudinal distances of 50.5 m and 86 m
from the spillway crest. On prototype scale, the maximum
absolute pressure difference was predicted at the longitudinal
distance of 86 m from the crest as ∆H = 19.2 m for Q10000 = 5053
m3/s. The possible source of error was considered from the
selection of uniform mesh size as ∆x = ∆y = ∆z = 0.5 m
throughout the computation domain. For a finer meshing with
nested mesh blocks (e.g. ∆x = ∆y = ∆z = 0.25 m or finer) better
predictions around baffle blocks could be expected from the
CFD model that will be part of the another subsequent research
study.
The data presented in Fig. 11 demonstrates that CFD modeling
is capable of reasonably predicting pressures on spillways and
stilling basins. The concern of modeling supercritical flow
transitioning to subcritical flow has been still a difficult problem
to solve, however numerical advances are rapidly reducing the
inherent difficulties of this problem (Savage and Johnson, 2006).
7. Conclusions
In this study, 1/50-scaled physical model was conducted in
order to investigate flow conditions and rating curves for full
openings of the radial gates of the spillway and flow over the
spillway for the operating conditions in the Kavsak Dam. A
series of experiments are tested in the State Hydraulic Works
Hydraulic Laboratory. Some modifications were done to obtain
uniform flow conditions and decreasing the energy losses
through the approach. Cavitation risk was tested flow along the
spillway. Aerators and damps are added as there was cavitation
risk. After observing final design for the approach flow
conditions and spillway, an attempt was made to simulate flow
over a spillway structure using commercially available CFD
software. Obtained results from the full-scaled (prototyped) CFD
model was compared to existing physical model data of the
Kavsak Dam and HEPP.
The flow rate results show that the CFD model provided a
reasonable solution. The average relative percent difference
between the CFD model and the physical model was obtained as
3.2%.
The CFD results obtained for free surface elevation and depth-
averaged velocity fit generally the physical model data, whereas
some difficulties observed at the flow transition from supercritical
to subcritical through the hydraulic jump region mainly due to
effects of high turbulence and flow bulking.
Although numerical methods offer a potential to provide
solutions with increasing accuracy, physical model studies are
still considered as the basis from which all other solution
methods used.
References
Bureau of Reclamation (1977). Design of small dams, U.S. Government
Printing Office, Washington, D.C., U.S.
Bureau of Reclamation (1990). Cavitation in chute and spillways,
Engineering Monograph, No.42, U.S.
Chanel, P. G. (2008). An evaluation of computational fluid dynamics for
spillway modeling, MSc Thesis, University of Manitoba Winnipeg,
Manitoba, Canada.
Chanson, H. (2002). The hydraulics of stepped chutes and spillways,
Balkema, Lisse, The Netherlands.
Chanson, H. and Gonzalez, C. A. (2005). “Physical modeling and scale
effects of air-water flows on stepped spillways.” Journal of Zhejiang
University Science, Vol. 6A, No. 3, pp. 243-250.
Demiroz, E. (1986). “Specifications of aeration structures which are
added to the spillways.” DSI Report, HI-754, DSI-TAKK Publications,
Ankara, Turkey.
Erfanain-Azmoudeh, M. H. and Kamanbedast, A. A. (2013). “Determine
the appropriate location of aerator system on gotvandoliadam's
spillway using Flow 3D.” American-Eurasian J. Agric. & Environ.
Sci., Vol. 13, No. 3, pp. 378-383, DOI: 10.5829/idosi.aejaes.2013.
13.03. 458.
Falvey, H. T. (1990). Cavitation in chutes and spillways, Engineering
Monograph 42 Water Resources Technical Publication US Printing
Office, Bureau of Reclamation, Denver.
Flow-3D User ’s Manual (2012). Flow science, Inc., Santa Fe, N.M.
Hirt, C. W. (1992). “Volume-fraction techniques: Powerful tools for flow
modeling.” Flow Science Report, No. FSI-92-00-02, Flow Science,
Inc., Santa Fe, N.M.
Hirt C. W. and Nichols B. D. (1981). “Volume of Fluid (VOF) method
for the dynamics of free boundaries.” Jornal of Computational
Physics, Vol. 39, pp. 201-225, DOI: 10.1016/0021-9991(81)90145-5.
Hirt, C. W. and Sicilian, J. M. (1985). “A Porosity technique for the
definition of obstacles in rectangular cell meshes.” Proceedings of
the 4th International Conference on Ship Hydro-dynamics, 24-27
September 1985, National Academic of Sciences, Washington DC.
Ho, D., Boyes, K., Donohoo, S., and Cooper, B. (2003). “Numerical
flow analysis for spillways.” 43rd ANCOLD Conference, Hobart,
Tas m a nia .
Johnson, M. C. and Savage, B. M. (2006). “Physical and numerical
comparison of flow over ogee spillway in the presence of tailwater.”
Journal of Hydraulic Engineering, Vol. 132, No. 12, pp. 1353-135,
DOI: 10.1061/(ASCE)0733-9429.
Kim, S. D., Lee, H. J., and An, S. D. (2010). “Improvement of hydraulic
stability for spillway using CFD model.” Int. Journal of the Physical
Sciences, Vol. 5, No. 6, pp. 774-780.
Kokpinar, M. A. and Gogus, M. (2002). “High speed jet flows over
spillway aerators.” Canadian Journal of Civil Engineering, Vol. 29,
No. 6, pp. 885-898, DOI: 10.1139/l02-088.
Kumcu, S. Y. (2010). Hydraulic model studies of Kavsak Dam and
HEPP, DSI Report, HI-1005, DSI-TAKK Publications, Ankara,
Serife Yurdagul Kumcu
−10 −KSCE Journal of Civil Engineering
Turkey.
Margeirsson, B. (2007). Computational modeling of flow over a spillway,
MSc Thesis, Chalmers University of Technology, Gothenburg, Sweden.
Nichols, B. D. and Hirt, C. W. (1975). “Methods for calculating multi-
dimensional, transient free surface flows past bodies.” Proc. First
Intern. Conf. Num., Ship Hydrodynamics, Gaithersburg, ML.
Savage, B. M. and Johnson, M. C. (2001). “Flow over ogee spillway:
Physical and numerical model case study.” Journal of Hydraulic
Engineering, ASCE, Vol. 127, No. 8, pp. 640-649, DOI: 10.1061/
(ASCE)0733-9429.
Souders, D. T. and Hirt, C. W. (2004). “Modeling entrainment of air at
turbulent free surfaces.” Critical Transitions in Water and Environmental
resources Management, pp. 1-10.
entürk, F. (1994). Hydraulics of dams and reservoirs, Water Resources
Publication Colorado, USA.
Teklemariam, E., Korbaylo, B, Groeneveld, J., Sydor, K., and Fuchs, D.
(2001). Optimization of hydraulic design using computational fluid
dynamics, Waterpower XII, Salt Lake City, Utah.
Teklemariam, E., Shumilak, B., Sydor, K., Murray, D., Fuchs, D., and
Holder, G. (2008). “An integral approach using both physical and
computational modeling can be beneficial in addressing the full
range of hydraulic design issues.” CDA Annual Conference,
Winnipeg, Canada.
Usta, E. (2014). Numerical investigation of hydraulic characteristics of
Laleli Dam spillway and comparison with physical model study,
Master Thesis, Middle East Technical University, Ankara, Turkey.
Versteeg, H. K. and Malalasekera, W. (1996). An introduction to
computational fluid dynamics, Longman Scientific and Technical,
Longman Group Limited, Harlow, England.
Vischer, D. L. and Hager, W. H. (1997). Dam hydraulics, J. Wiley &
Sons Ltd., England.
Wagner, W. E. (1967). “Glen Canyon diversion tunnel outlets.” J.
Hydraulic Division, ASCE, Vol. 93, No. HY6, pp. 113-134.
Willey, J., Ewing, T., Wark, B., and Lesleighter, E. (2012). Comple-
mentary use of physical and numerical modeling techniques in
spillway design refinement, Commission Internationale Des Grands
Barrages, Kyoto, June 2012.
S
ç
