Journal of Groundwater Science and Engineering Vol.7 No.

4: 323—332

DOI: 10.19637/j.cnki.2305-7068.2019.04.003

Numerical investigation of hydraulic characteristics and

prediction of cavitation number in Shahid Madani Dam's
Spillway
Babak Ghazi1, Rasoul Daneshfaraz2*, Esmaeil Jeihouni3
1
Department of Civil Engineering, Sofian Branch, Islamic Azad University, Sofian, Iran.
2
Department of Civil Engineering, Faculty of Engineering, University of Maragheh, Maragheh, Iran.
3
Department of Civil Engineering, Sofian Branch, Islamic Azad University, Sofian, Iran.

Abstract: The prediction of the probability of cavitation occurrence to prevent serious

damages in the spillways is the major concern for hydraulic engineers. In this research,
the three-dimensional model of Shahid Madani Dam's spillway was simulated with the
Flow 3D software and by the comparison of numerical model results with the experi-
mental data, the probability of occurrence of the cavitation phenomenon has been
investigated. The flow parameters including pressure, velocity, and water depth were
calculated for three different flow rates of 495 m3/s, 705 m3/s and 2 205 m3/s respectively.
The Renormalization Group (RNG) turbulence model was used to simulate current turbu-
lence. Comparison of simulation results for pressure, velocity and water depth with the
results of the experimental model with two statistical indices Root Mean Square Error
(RMSE) and Coefficient of Determination (R2) showed that the numerical simulation
results are in good consistency with experimental model. However, simulation results indi-
cated that at any flow rates with a return period of 1 000 years, probable maximum flood
and designed flow rates, the cavitation number is not lower than the critical cavitation
number. Therefore, it is predicted that the cavitation phenomenon in Shahid Madani
Dam's spillway will not happen.

Keywords: Spillway; Cavitation index; Velocity; Water depth; Pressure distribution

Received: 28 Jun 2019/ Accepted: 19 Aug 2019

Introduction red to spillways of Glen Canyon, Blue Mesa,

Yellowtail and Hoover dams in the United States
Cavitation is one of the most complicated and as well as the concrete surface of the Shahid
destructive phenomena that occur in the spillways. Abbaspour Dam's spillway in Iran has worn out
Increasing velocity due to flow passage through due to cavitation. Since the cavitation phenomena
the spillway could lead to low pressure and it may may heavily damage the spillway, it can be a
cause cavitation, leading damage of surface con- major threat to dam safety. Therefore, it is very
crete in the spillway (Falvey Henry T, 1990). important to prevent the occurrence of hydraulic
It is well documented that there are examples cavitation in the spillways.
of cavitation damage in dam spillways in the The prediction and simulation of the cavitation
world. For instance, cavitation damage that occur- phenomenon in the spillways have a significant
impact on the optimal designing and protecting in
*
Corresponding author: [email protected]
the spillway. Nevertheless, the prediction of cavita-
http://gwse.iheg.org.cn 323
Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

tion in spillways is difficult due to uncertain condi- spillway. The result of simulation for aeration in
tions on the structure, complexity of this pheno- spillway shows that there is a suitable agreement
menon and its various flow conditions combining between Fluent software and experimental data
with the reaction of the surface material (Dular which proves the aeration simulation results in the
Matevž et al. 2006). numerical models by comparing with experimen-
Generally, engineers use prototype and experi- tal models are acceptable. Vosoughifar et al.
mental models to forecast cavitation. Since experi- (2013), by using a CFD code-named V-Flow in
mental and prototype models are an expensive and MATLAB software, investigatied unstable flow
time-consuming process, with the development of over spillway. The simulation results of MATLAB
various Computational Fluid Dynamic (CFD) V-flow code for velocity, static pressure, and total
models, using numerical models to predict and pressure were validated by Fluent software results,
investigate hydraulic characteristics and cavitation then by comparing results of two numerical mo-
phenomenon is a reasonable method (Johnson dels with the experimental model, indicating that
Michael C and Bruce M Savage, 2006). there is reasonable agreement between V-flow,
Over the past years, numerous experimental fluent and experimental model. Daneshfaraz R et
and numerical studies were conducted on spill- al. (2014) studied two numerical models of flow
ways, among which the most important researches over ogee and stepped spillways in Fluent soft-
are as follows. Olsen et al. (1998) used 2D and 3D ware with Finite Volume method and ADINA
numerical models to predict water flow and spill- software with the Finite Element method. The con-
way capacity over ogee spillway. Reynolds Ave- ducted simulation results for ADINA and Fluent
raged Navier Stokes (RANS) Equations were software show that there are good agreements
solved with the K-epsilon (k-ε) Turbulence Model between two numerical models and experimental
in Flow 3D software. Results of numerical models model. However, in some stations of the spillway,
compared with the experimental study showed that the accuracy of Finite Volume method of Fluent
there was a good agreement on discharge coe- software results is more acceptable than the Finite
fficient and the pressure distribution on the spill- Element of ADINA software results. Parsaie et al.
way and numerical models results were close to (2016) conducted flow parameters and cavitation
the physical model results. Savage et al. (2001) phenomenon along with the spillway's flip bucket
investigated flow parameters over a standard ogee- of the Balaroud Dam by simulation Flow 3D nu-
crested spillway by applying once, the experimen- merical model and experimental model. Flow 3D
tal model and then with the numerical simulation. model results shows that the RNG model has been
In the experimental model, they recorded the flow a suitable performance for hydraulic parameters
and pressure for different flow conditions. Then, and cavitation index. Bayon et al. (2018) studied
simulation of flow over ogee spillway in the Flow with numerical modeling of water flow in stepped
3D software with the volume of fluid (VOF) and spillway without aeration by Flow 3D and Open-
fractional area-volume obstacle representation Foam models. They used RNG and k-ε methods to
(Favor) models were examined in the pressure. simulation turbulent flow. It was found that the
The study shows that numerical tools were suffi- result of the k-ε method was very similar to experi-
ciently advanced to calculate discharge and pre- mental model results and there was only a 3%
ssure on the spillway. Chatila et al. (2004) mo- difference between RNG method and experimental
deled the water flow over ogee spillway with k- ε model. WAN Wu-yi et al. (2018) studied a nu-
model in ADINA software which is a Finite Ele- merical simulation over ogee spillway in Fluent
ment Fluid Mechanics simulation. The results indi- software. The comparison of Fluent model results
cate that the ADINA software has an acceptable with experimental model results stated that Fluent
performance on simulation of hydraulic characte- software is able to predict cavitation number in
ristics, and the compared numerical model has a spillways and it has an economical approach and it
good agreement with experimental results. Aydin is also a cost-effective option. Furthermore, Fluent
et al. (2009) used k-ε and RNG models in Fluent could give complete information about the prob-
software and empirical Equations and previous ability of cavitation phenomenon in different sec-
studies to investigate the effect of aeration on the tions of the spillway.

324 http://gwse.iheg.org.cn
 Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

According to literatures, results from studies by utilizing Flow 3D software.

which employed Flow 3D for modeling of various
spillways indicate good agreement with data
obtained from experimental model studies. Simul- 1 Material and methods
taneously, Flow 3D can successfully model water
surface profile, flow velocity and pressure distri- The Shahid Madani Dam is an earth-fill dam
bution for a variety of headwater levels and diffe- with clay core located at Vaniyar city, 8 km far
rent gate openings as compared to physical model from Tabriz (East-Azerbaijan Province) in Iran.
testing in spillways (Chanel Paul Guy, 2009). The The Shahid Madani Dam has an ogee spillway
advantage of the Flow 3D model in comparison which 1 000 years, designed and the probable
with other models in computational fluid dyna- maximum flood flow rates are 495 m3/s, 705 m3/s
mics is that it has a user-friendly and powerful and 2 205 m3/s respectively. The general features
graphical interface (Taghavi Mehdi and Hesam and geographical position of Shahid Madani Dam
Ghodousi, 2015). Therefore, in present research, are shown in Table 1 and Fig. 1 (Iranian Water
the hydraulic parameters and cavitation number in Research Institute, 2009).
the Shahid Madani Dam's spillway were simulated

Table 1 Features of Shahid Madani Dam

Dam type Height (m) Length (m) Crest width (m) Reservoir capacity (m3)
Earth-fill dam with silt-clay core 44 257 10 363

Fig. 1 Geographical position of Shahid Madani Dam

The experimental model of Shahid Madani (Iranian Water Research Institute, 2009).
Dam's spillway was designed and constructed in In this research, Flow 3D software was used to
Water Resources Research Institute of the Mini- investigate the hydraulic characteristics and cavita-
stry of Energy on the scale of 1:40. Fig. 2 illustra- tion number in Shahid Madani Dam's spillway.
tes the scaled experimental models. In the experi- First, the hydraulic parameters of flow over the
mental model, the values of pressure, velocity, and spillway were calculated by numerical simulation.
water depth were measured in numerous sections. The numerical simulation results were compared
The velocity of the spillway was measured using a and validated by experimental model results. Then
pitot tube and measurement accuracy was ± 0.14 the cavitation number will be predicted on the
m/s. A precise limn meter was used to measure the spillway.
water depth with an accuracy of ± 0.1 mm and After numerical simulation of flow over the
values of pressure measurement were performed spillway, the measurements of water depth, velo-
by a piezometer with the accuracy of ± 0.1 mm city, and pressure were calculated at eight sections

http://gwse.iheg.org.cn 325
Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

Fig. 2 Experimental models of Shahid Madani Dam's spillway (Iranian Water Research Institute, 2009)

over the spillway. The locations of the selected sections are presented in Fig. 3 and Table 2.

Fig. 3 Locations of the selected sections

Table 2 Features of selected sections

Section name A B C D E F G H
Distance from spillway crest (m) 147.0 157.1 169.7 182.8 195.6 211.6 218.0 224.4

2 Numerical model ∂w
∂t
+
1
VF ( uA x
∂w
∂x
+ vA y
∂w
∂y
+ wA z
∂w
∂z
= )
The Flow 3D is one of the most complete, 1 ∂k
- + Gz + fz (4)
powerful and practical CFD softwares for simula- ρ ∂z
tion various flows. The software for solving three- The Equations are written for the fluid velocity
dimensional Reynolds Averaged Navier-Stokes in three Cartesian coordinate systems, where
Equations utilizes the finite volume method (Flow- A: The fractional areas associated with the
Science, 2014). In the three Cartesian coordinate flow. G: Local accelerations of the fluid. f: Fric-
systems, these Equations can be written as follows: tional forces in the respective directions. ρ: Signi-
∂ρ ∂ (ρuA x ) ∂ (ρvA y ) ∂ (ρwA z ) fies the fluid density. k: Local pressure.
VF + + + = R SOR
∂t ∂x ∂y ∂z The Flow 3D software uses Volume of Fluid
(1) for the free surface flow models and Favor techni-
∂u
∂t
+
1
VF( uA x
∂u
∂x
+ vA y
∂u
∂y
+ wA z
∂u
∂z
= ) que utilizes for models with complex geometry
(Flow-Science, 2014). This method is showed as
1 ∂k follows:

(
- + Gx + fx (2)
ρ ∂x ∂F 1 ∂ ∂
+ ( FuA x ) + (FvA y ) +
∂v
∂t
+
1
VF (
uA x
∂v
∂x
+ vA y
∂v
∂y
+ wA z
∂v
∂z
= ) ∂t V f ∂x
∂
(FwA z ) = 0)
∂y

(5)
1 ∂k ∂z
- + Gy + fy (3)
ρ ∂y where F is the fraction function, F=0 when no
326 http://gwse.iheg.org.cn
 Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

water inside and F=1 when the cell is full of water. 3 Boundary conditions
There are various well-known models in Flow
3D software for simulation turbulence flows such In order to define the boundary conditions, in
as k- ε, RNG and large eddy simulation model the inflow point (Ymin) volume flow rate, in the
(Daneshfaraz Rasoul et al. 2019). In this research, sides of the spillway (Xmin, Xmax) and bed wall con-
the RNG model was used to simulate the Shahid dition, on the upper border (Zmax) symmetry condi-
Madani Dam's spillway. The base of this model is tion and in the outflow section of the spillway,
k-ε which was developed by Jones W P and Brian outflow condition was used. The boundary condi-
Edward Launder (1972), and Launder Brian Ed- tions of Shahid Madani Dam's spillway in the
ward and Dudley Brian Spalding (1972). The final Flow 3D software were shown in Fig. 4 and Table
adjustment was presented by Yakhot VSASTBCG 3 respectively.
et al. (1992). The geometry of spillway was
created in AutoCAD 3D 2013 software then STL
file format was imported to Flow 3D.

Fig. 4 Applied boundary condition in the Flow 3D

Table 3 Detail of boundary condition in Flow 3D

Free surface Lateral
Software Upstream boundary Downstream boundary Floor boundary
boundary boundary
FLOW 3D Volume flow rate Out flow Symmetry Wall Wall

The determination of suitable type and size of were analyzed at pressure in the 2 205 m3/s flow
the grid has a significant impact on simulation rates to determine appropriate mesh models. Fig. 5
results. The grid with large sizes gives an inappro- shows different mesh models for Shahid Madani
priate result. By increasing the number of meshes, Dam's spillway in Flow 3D software.
the time of simulation will be increased but the According to the compared different mesh
simulation results will be close to experimental models in Fig. 5. The results of mesh Types 3 and
results. However, after a certain amount of mesh 4 are close to experimental model results. Never-
number, the result will not be changed by mesh theless, the time of simulation in Type 3 is more
size. In this research, four different mesh models appropriate than that of Type 4, so Type 3 has

http://gwse.iheg.org.cn 327
Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

Fig. 5 Different mesh models based on pressure

been chosen for simulation and all numerical re- at G and H points. The highest velocity at Shahid
sults were presented by this mesh type. The num- Madani Dam's spillway is 25.3 m/s, 28.4 m/s and
ber of cells for different mesh types is shown in 28.8 m/s for the three flow rates of 495 m3/s, 705
Table 4. m3/s and 2 205 m3/s respectively. It can be seen
that, with an increases in flow rate, the amount of
Table 4 Mesh type and the number of cells velocity increase over the spillway. The simulation
Mesh model namel Number of cell results for the pressure indicate that the pressure
Type 1 442 000
values reduce over the spillway and the lowest
amount of pressure over the spillway occurs at 495
Type 2 584 000
m3/s flow rates, which is 0.4 m. However, the
Type 3 830 000
numerical values for the water depth also indicate
Type 4 1 096 000
that the water depth in 2 205 m3/s flow rates peaks
at 8.4 m, which is the maximum water depth in all
three flow rates. Finally, the minimum water depth
4 Results and discussion appears in 495 m3/s flow rates at 0.7 m.
To determine the difference between the expe-
By numerical simulation in Flow 3D, hydrau- rimental model and numerical model, Root Mean
lic characteristics, including flow depth, velocity, Square Error (RMSE) and the Coefficient of Deter-
and pressure for three flow rates 495 m3/s, 705 mination (R2) were used. For a reasonable fit
m3/s and 2 205 m3/s were calculated and compared between numerical models and experimental ones,
with experimental models results respectively. The the amount of R2 should be close to 1, and the
results of computed numerical models were com- values of RMSE close to zero. The amount of
pared with the experimental model results for flow RMSE and R2 for different hydraulic parameters in
depth, the pressure, and velocity of Shahid Madani Shahid Madani Dam's spillway shows that there is
Dam's spillway and shown in Fig. 6-Fig. 8. acceptable precision for the numerical and experi-
Given that the spillway has the longitudinal mental model results. The values of RMSE and R2
slope, as shown in Fig. 6, the velocity increases are shown in Table 5.
along the path and it reaches to the highest amount

328 http://gwse.iheg.org.cn
 Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

Fig. 6 Numerical and experimental model results for velocity

Fig. 7 Numerical and experimental model results for water depth

Table 5 Result of RMSE and R2 5 Cavitation number

Statistical indices Flow depth Velocity Pressure
RMSE 0.0564 0.0342 0.0769 The occurrence of cavitation on the spillway
can be obtained by using a unit which is called
R2 0.891 0.976 0.981
cavitation number (Equation 6). To prevent spill-
way from cavitation damage, it is necessary to the
keep cavitation number more than the critical

http://gwse.iheg.org.cn 329
Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

Fig. 8 Numerical and experimental model results for pressure

cavitation number. However, it is predicted that cavitation damage in

P - Pv Shahid Madani Dam's spillway will not occur and
σ=
V2 (6) the spillway is not required to the aeration system.
ρ
2 Fig. 9 presents the cavitation number for 495 m3/s,
where: P is a reference for pressure; Pv is water 705 m3/s and 2 205 m3/s flow rates. The results of
vapor pressure and ρ is water density. the experimental model and numerical model of
To predict cavitation damage in spillways, Shahid Madani Dam's spillway are investigated
there are four different levels which considered the and compared. As regards the several years of
risk of cavitation on a spillway (Falvey Henry T, exploitation of the Dam and the spillway of the
1990), as shown in Table 6. Shahid Madani, so far the cavitation phenomenon
has not occurred in the spillway of the Shahid
Table 6 Level of cavitation damage Madani Dam. Therefore, the investigation of nu-
Cavitation merical and experimental models confirms what
Level Damage risk
number had happened in the real Shahid Madani Dam's
1 σ>0.2 No cavitation damage spillway. Unfortunately, the results of real Shahid
2 σ˂0.2 Possible cavitation damage Madani Dam's project are not available for more
investigation.
3 0.1˂σ˂0.2 The aeration system is necessary
Spillway should be designed
4 σ˂0.1
again
6 Conclusions
According to the result for cavitation number
for three flow rates of 495 m3/s, 705 m3/s and In this research, the flow over the Shahid
2 205 m3/s, the lowest amounts for cavitation num- Madani Dam's spillway was simulated by using
ber are 0.31, 0.29 and 0.326 respectively. With Flow 3D software. The flow depth, velocity, and
regards to Table 6 the result of cavitation number pressure for three different flow rates were calcu-
for Shahid Madani Dam's spillway indicates that, lated. Computing results of RMSE and R2 indicate
in non-sections of spillway, cavitation numbers are that the numerical model is consistent with the ex-
not lower than the critical cavitation number (0.1). perimental model, so the Flow 3D software can

330 http://gwse.iheg.org.cn
 Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

Fig. 9 Cavitation number for Shahid Madani Dam's spillway

accurately simulate the hydraulic characteristics Daneshfaraz R, B Kaya, S Sadeghfam, et al. 2014.
and cavitation number. Simulation of flow over ogee and stepped
The obtained cavitation numbers for the 495 spillways and comparison of finite element
m /s, 705 m3/s and 2 205 m3/s flow rates by velo-
3 volume and finite element methods. Journal
of Water Resource and Hydraulic Engineer‐
city and pressure distribution show that the cavita-
ing, 3(2): 37-47.
tion numbers are higher than the critical cavitation
number in the specified sections, so it is estimated Daneshfaraz Rasoul, Omar Minaei, John Abra‐
that cavitation phenomenon in Shahid Madani ham, et al. 2019. 3-D numerical simulation of
water flow over a broad-crested weir with
Dam's spillway will not occur for the three men-
openings. ISH Journal of Hydraulic Engineer‐
tioned flow rates. ing: 1-9.
References Dular Matevž, Bernd Stoffel, Brane Širok. 2006.
Development of a cavitation erosion model.
Aydin M Cihan, Mualla Ozturk. 2009. Verification
Wear, 261(5-6): 642-655.
and validation of a computational fluid dy‐
namics (CFD) model for air entrainment at Falvey Henry T. 1990. Cavitation in chutes and
spillway aerators. Canadian Journal of Civil spillways. Engineering Monograph 42. Water
Engineering, 36(5): 826-836. Resources Technical Publication. US Printing
Office. Bureau of Reclamation, Denver.
Bayon Arnau, Juan Pablo Toro, Fabián A Bom‐
bardelli, et al. 2018. Influence of VOF tech‐ Flow-Science. 2014. FLOW-3D user manual, ver‐
nique, turbulence model and discretization sion 11. In: Flow Science Santa Fe, NM.
scheme on the numerical simulation of the
non-aerated, skimming flow in stepped spill‐ Iranian Water Research Institute. 2009. Shahid
ways. Journal of Hydroenvironment Re‐ Madani Dam's spillway final report. Ministry
search, 19:137-149. of Energy (Tehran Iran): C8202-PR.

Chanel Paul Guy. 2009. An evaluation of computa‐ Johnson Michael C, Bruce M Savage. 2006. Physi‐
tional fluid dynamics for spillway modeling. cal and numerical comparison of flow over
ogee spillway in the presence of tailwater.
Chatila Jean, Mazen Tabbara. 2004. Computation‐ Journal of Hydraulic Engineering, 132(12):
al modeling of flow over an ogee spillway. 1353-1357.
Computers & Structures, 82(22): 1805-1812.
http://gwse.iheg.org.cn 331
Journal of Groundwater Science and Engineering Vol.7 No.4: 323—332

Jones W P, Brian Edward Launder. 1972. The pre‐ Taghavi Mehdi, Hesam Ghodousi. 2015. Simula‐
diction of laminarization with a two-equation tion of flow suspended load in weirs by using
model of turbulence. International Journal of flow 3D model. Civil Engineering Journal, 1
Heat and Mass Transfer, 15(2): 301-314. (1): 37-49.
Launder Brian Edward, Dudley Brian Spalding. Vosoughifar Hamid Reza, Azarri Dolatshah, Seyed
1972. Mathematical models of turbulence Kazem Sadat Shokouhi, et al. 2013. Evalua‐
(Academic Press). tion of fluid flow over stepped spillways us‐
ing the finite volume method as a novel ap‐
Olsen Nils R B, Hilde M Kjellesvig. 1998. Three- proach. Strojniski Vestnik/ Journal of Me‐
dimensional numerical flow modelling for es‐ chanical Engineering, 59(5): 301-310.
timation of spillway capacity. Journal of Hy‐
draulic Research, 36(5): 775-784. WAN Wu-yi, LIU Bin, Awais Raza. 2018. Numeri‐
cal prediction and risk analysis of hydraulic
Parsaie Abbas, Sadegh Dehdar-Behbahani, Amir cavitation damage in a High-Speed-Flow
Hamzeh Haghiabi. 2016. Numerical model‐ spillway. Shock and Vibration.
ing of cavitation on spillway's flip bucket.
Frontiers of Structural and Civil Engineering, Yakhot VSASTBCG, SA Orszag, S Thangam, et
10(4): 438-444. al. 1992. Development of turbulence models
for shear flows by a double expansion tech‐
Savage Bruce M, Michael C Johnson. 2001. Flow nique. Physics of Fluids A: Fluid Dynamics, 4
over ogee spillway: Physical and numerical (7): 1510-1520.
model case study. Journal of Hydraulic Engi‐
neering, 127(8): 640-649.

332 http://gwse.iheg.org.cn