11th World Congress on Computational Mechanics (WCCM XI)

5th European Conference on Computational Mechanics (ECCM V)

6th European Conference on Computational Fluid Dynamics (ECFD VI)
E. Oñate, J. Oliver and A. Huerta (Eds)

COMPUTATION OF LAMINAR AND TURBULENT WATER

HAMMER FLOWS

S. D. SAEMI1, 2, M. RAISEE1, 2, M. J. CERVANTES3 AND A. NOURBAKHSH1, 2

1
Hydraulic Machinery Research Institute (HMRI), School of Mechanical Engineering
College of Engineering, University of Tehran, P. O. Box: 11155-4563, Tehran, Iran.
2
Center of Excellence in Design and Optimization of Energy Systems (CEDOES)
School of Mechanical Engineering, College of Engineering, University of Tehran,
P.O.Box: 11155-4563, Tehran, Iran.
[email protected]
3
Division of Fluid and Experimental Mechanics,
Luleå University of Technology, SE 971 87, Lulea, Sweden.
[email protected]
Key Words: Water hammer, laminar flow, turbulent flow, numerical simulation.

Abstract. In this paper, the water hammer phenomenon in a pipeline is simulated using the
full Reynolds-Averaged Navier-Stokes equations. The flow is considered to be compressible
and the effect of pipe elasticity is taken into account by introducing the bulk modulus of
elasticity in the solution procedure. Computations are performed both for laminar and
turbulent flows. The high-Re RNG k-ε and the low-Re k-ω SST turbulence models are
employed for turbulence modeling. Numerical results for both laminar and turbulent flows are
compared with the available experimental data and numerical results in the literature. For the
laminar flow test case, the head variation shows good agreement with the experimental data.
Comparisons for turbulent test case show that the RNG k-ε model somewhat overpredicts the
head variation. The low-Re k-ω SST model, in the other hand, produces more accurate wall
shear stress distribution than the high-Re RNG k-ε model. This highlights the importance of
implementation of low-Re turbulence models for the prediction of water hammer flows.

1 INTRODUCTION

The term ‘water hammer’ is used to describe the fluid flow characteristics when a fluid in
motion is forced to stop in closed systems such as a pipe network. This phenomenon produces
intense pressure waves that travel periodically along the pipe. Water hammer mostly occurs in
piping systems, e.g., in power plants and urban water carrier systems, due to a sudden change
in the flow rate during a sudden closure (or opening) of a valve or pump failure involving a
density variation. Many researchers have used experimental as well as theoretical methods to
examine the phenomena of the water hammer in straight pipes. An example of experimental
works is the study of Holmboe et al. [1] who measured the pressure variation in transient
laminar flow and compared the results with frictionless analysis of Joukowski [3]. It was
shown that the theoretical solution considering inviscid flow is valid when low viscous liquid
like water is employed in the measurements. Bergant et al. [2] measured the pressure
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

variation, caused by fast closure of the valve, in laminar and low Reynolds turbulent flows.
Safwat [6] measured the strains in the pipe wall, resulting from fast closure of the valve, to
evaluate the elastic behavior of the pipe during the water hammer. They concluded that, the
strain measurement on the outer side surface of the pipe wall can yield indirectly the transient
pressure changes in the pipe. Brunone et al. [7] measured pressure and velocity profiles
caused by water hammer and compared the measured pressure change with the results of 1D
simulation. Concerning analytical studies, one can refer to the work of Joukowsky [2] who
proposed the following well-known formula for the piezometric pressure which is called
‘fundamental equation of water hammer’.
a∆u
∆H= ± (1)
g

where a , H and u are respectively the wave speed, the piezometric head and the cross
sectional average velocity. Ghidaoui and Kolyshkin [8] performed a linear stability analysis
for the velocity profiles in both unsteady laminar and turbulent flows caused by water
hammer. They demonstrated that the sources of flow instability are the presence of inflection
points in the velocity profile and large gradient near the pipe wall.
One of the main issues in water hammer modeling is the application of reliable unsteady
friction models for predictions. Therefore, a number of researches performed experimental
and analytical investigations to develop accurate unsteady friction models. Among the
proposed models for 1D simulation of water hammer, the Darcy-Weisbach shear stress for
head losses in turbulent pipe flow is the most well-known model. This model is based on
steady axial averaged velocity but remains valid under unsteady condition [9].
uu
τw = ρ f (2)
8
As reported in Ghidaoui et al. [10], the application of this model in very slow transient flows
(quasi-steady) gives acceptable results. However, water hammer flow occurs in very fast
transient conditions. To address this issue, the unsteady wall shear stress τ wu is introduced as:
τ wu = τ w −τ ws (3)

The unsteady wall shear stress has been modeled using empirical-based corrections on quasi-
steady shear stress models derived from laboratory experiments by Brunone et al [5],
Pezzinga [11] and Bergant et al [2] as well as physically based models which are based on
analytical solutions of the unidirectional flow used in 1D water hammer simulations. Among
the physically based models, the Zielke model [4] derived from the Laplace transform of the
axial component of the Navier-Stokes equations in laminar regime, is the most well-known
one. Bergant et al [2] and Trikha [12] employed Zielke model [4] to simulate turbulent
transient pipe flows and found that this model is applicable for low Reynolds turbulent
transient flows. Also the Vardy and Brown model [13, 14] is widely used for turbulent
transient flows in smooth and rough pipes by frozen turbulence model assumption. Due to the
limitations in obtaining detailed information in 1D simulations, e.g. velocity and pressure
fields and energy dissipation, 2D simulation using appropriate turbulence models should be
performed. The most widely used turbulence models in the literature are algebraic turbulence

2
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

models due to their simple mathematical formulation. Examples of application of algebraic

turbulence models for the prediction of water hammer problem are two-layer turbulence
model of Cebeci-Smith [15], five-region turbulence model of Kita et al. [16] and Baldwin and
Lomax [17]. More recently, Riasi et al. [24] used the Wilcox k-ω model to study water
hammer in a pipe.
Concerning the numerical methods used for the water hammer simulations, the most widely
used numerical scheme is the method of characteristics (MOC) introduced by Wylie et al. [25]
and Chaudhry [26] due to its simple numerical implementation and efficiency as well as its
ability to handle complex boundary conditions. This method transforms partial differential
equations into ordinary differential equations along characteristics lines. Integrating from one
time step to the next requires pressure and velocity to be known along each characteristics
line. As shown by Tijsseling et al. [27], the MOC method gives an accurate prediction of the
maximum pressure in the system which usually occurs during the first pressure peak. It also
correctly predicts wave periods, though it usually fails in the prediction of damping and
dispersion of wave fronts accurately. Chaudhry and Husssaini [28] used MacCormack,
Lambda and Gabutti explicit finite difference schemes to solve the water hammer equations.
Their work shows that the second order finite difference method (in time and space) produces
more accurate results than the first order MOC with courant number less than one. Another
important numerical method in the solution of water hammer hyperbolic system of equations
is the finite volume method. This method has some profits such as the conservative character
and the ability to provide the correct resolution of discontinuities resulting from the pressure
changes. This method was first used by Guinot [29] who employed the two-dimensional
Riemann solver. Moreover, Zhao and Ghaidaoui [30] implemented first- and second-order
explicit finite volume Godunov-type solution for water hammer simulation. They indicated
that the second order methods produce more accurate results than the first order ones. All of
the studies described in the literature relied on simplified water hammer equations for
predictions and did not use full Navier-Stokes equations.
For more precise prediction of the water hammer phenomena in complex 3D geometries, one
requires to employ full governing equations. For this purpose, in this work, we investigated
the water hammer in a straight pipe based on unsteady Reynolds averaged Navier-Stokes
(URANS) equations using finite volume method and without any change in the nature of the
governing equations. The CFD-based simulations of water hammer are performed using
FLUENT software. The simulations are conducted for both laminar and turbulent flows. The
high-Re RNG k-ε and the low-Reynolds k-ω SST turbulence models are employed for the
predictions. The predicted pressure head at the pipe centreline midpoint are compared with
the available experimental data.

2 PROBLEM DESCRIPTION AND MATHEMATICAL FORMULATION

2. 1 Test cases
The geometry considered for the water hammer simulations is shown in Fig. 1 and the details
of experiments are summarized in Table 1. As indicated in Table 1 computations of the
present work have been carried out for both laminar and turbulent water hammer flows. The
experimental work for the laminar test case was performed at Reynolds number of 82 by
Holmboe et al [1]. In order to get water hammer, a fast closing valve is located at the end of a

3
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

1" ID copper tube. The pipe is placed in concrete to damp the vibration created by the water
hammer. A 60 gallons capacity tank is utilized to have an infinite reservoir at the other end of
the tube. The tank pressure maintained constant by the use of compressed air and pressure
transducers were used to measure the head change in various locations. As described in
Bergant et al. [2], the turbulent experimental test case contains a flexible laboratory apparatus
involving a long copper pipe of 22.1 mm in diameter connected to two pressurized tanks. The
pressure of the two tanks was specified and controlled by computerized pressure control
system. The water hammer in the apparatus was initiated by the rapid closure of a ball valve.
In both experimental works the pressure change at the end and in the middle of the pipe was
measured.

Fig. 1: Pipe geometry used for the water hammer simuations.

The numerical simulations were performed with a 2D geometry shown in Fig. 1 assuming an
axis-symmetric flow.
Table 1: The details of the test cases
Pipe Wave Initial Dynamic Reynolds
Pipe length Density
Test No. diameter 3 velocity velocity viscosity Number
(m) (kg/m )
(m) (m/s) (m/s) (N s/m2) (-)
1 36.09 0.025 878.4 1324 0.13 0.03483 82
2 37.2 0.022 998.2 1276 0.3 0.0011545 5700

2. 2 Governing equations

For compressible laminar and turbulent flows, the conservation laws of mass and momentum
may be written as follows:
Continuity:
∂ρ ∂ ( ρ u j )
+ =0 (4)
∂t ∂x j

Momentum:

∂ ( ρ ui ) ∂ ( ρ u j ui ) ∂P ∂ ⎛ ∂u ⎞
+ =− + ⎜⎜ µ i − ρ ui′u ′j ⎟⎟ (5)
∂t ∂x j ∂xi ∂x j ⎝ ∂x j ⎠

It should be noted that the Reynolds stress term in Eq. (5) vanishes in the laminar flow. To
account for the effect of pipe elasticity a new Bulk modulus of elasticity K ′f = K f (1 + K f D eE )
is introduced. A rigid pipe assumption will give rise to a change of pressure larger than for an
elastic one. This leads to a higher wave speed which is incompatible with the experimental

4
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

data. Introducing a new Bulk modulus based on the change of pipe diameter is appropriate to
solve the water hammer phenomenon. The pressure wave speed is thus given by:
Kf ρ
a= (6)
1 + K f D eE

Where K f is the bulk modulus of elasticity of the fluid, E is the Young modulus of elasticity,
e is the thickness of the pipe, D is the diameter of the pipe and ρ is the density of the fluid.
The flow is assumed to be compressible and isothermal. Thus, the density variation depends
only on the pressure change and is computed based on the following equation:
d ρ = ρ dp / K ′f (7)

This equation provides the relationship between the pressure and the density change and
implemented by a User Defined Function (UDF) into the FLUENT software.

2. 3 Turbulence modeling equations

In order to simulate turbulent water hammer, the high-Re RNG k-ε and the low-Re k-ω SST
turbulence models were used for the computations. The mathematical formulation of the RNG
k-ε model is similar to the standard k-ε but with some modifications [31]. The wall functions
approach is used to simulate the flow with the RNG k-ε model. The k-ω SST model is suitable
for the prediction of turbulent flows with the pressure gradient

2. 3. 1 The high-Re RNG k-ε turbulence model

In the high-Re RNG k-ε turbulence model, the Reynolds stress tensor in the momentum Eq.
(5) is modeled by the eddy viscosity approximation:
⎛ ∂u ∂u j ⎞ 2
−ui′u ′j = ν t ⎜ i + ⎟⎟ − δ ij k (8)
⎜ ∂x
⎝ j ∂xi ⎠ 3

where k is the turbulent kinetic energy and ν t is turbulent kinematic viscosity, which is
defined as:
k2 (9)
ν t = Cµ
ε
To obtainν t , the following transport equations are solved for the turbulent kinetic energy and
its dissipation rate:
∂ ∂ ∂ ⎛⎛ µ ⎞ ∂k ⎞
( ρ k ) + ( ρ ku j ) = ⎜⎜ ⎜ µ + t ⎟ ⎟⎟ + Pk − ρε (10)
∂t ∂x j ∂x j ⎝ ⎝ σ k ⎠ ∂x j ⎠
∂ ∂ ∂ ⎡⎛ µ ∂ε ⎞ ⎤ εP ε2
∂t
( ρε ) +
∂x j
( ρu jε ) = ⎢⎜ µ + t
∂x j ⎣⎢⎜⎝
⎟ ⎥ + Cε 1 k − Cε*2 ρ
σ ε ∂x j ⎟⎠ ⎥⎦ k k
(11)

5
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

where Pk is the turbulent energy production. C2*ε = C2ε + Cµη 3 (1 −η η0 ) (1 + βη 3 ) and η = S k ε

S = ( 2 S ij S ij )
1/ 2
with . All model constants are given in Table 2.
Table 2: Empirical constants for the k-ε model
Cµ Cε 1 Cε 2 σk σε η0 β
0.0845 1.42 1.68 0.7194 0.7194 4.38 0.012

2. 3. 2 The low-Re k-ω SST model

In the low-Re k-ω SST model, the unknown Reynolds stress tensor is also obtained from the
Boussinesq approximation (Eq. 8). The k-ω SST model uses the original k-ω model of Wilcox
in the inner region and the standard k-ε in the outer region of the boundary layer which
overcomes the strong sensitivity of the Wilcox model to the free stream conditions [32]. In
this model the eddy viscosity ν t is obtained from:
ν t = a1k max ( a1ω , ΩF2 ) (12)

where Ω is the absolute value of vorticity. Moreover, the turbulent kinetic energy, k, and its
specific dissipation, ω, are obtained from the following equations:
∂ ∂ ∂ ⎛ ∂k ⎞
( ρ k ) + ( ρ ku j ) = ⎜⎜ ( µ + σ k1µt ) ⎟⎟ + Pk − β ρ kω
*
(13)
∂t ∂x j ∂x j ⎝ ∂x j ⎠
∂ ∂ ∂ ⎡ ∂ω ⎤ γ
∂t
( ρω ) +
∂x j
( ρ u jω ) = ⎢( µ + σ ω1 µt )
∂x j ⎢⎣
⎥ + Pk
∂x j ⎥⎦ ν t
(14)
1 ∂k ∂ω
− βρω 2 + 2 (1 − F1 ) ρσ ω 2
ω ∂x j ∂ x j
⎧⎧ ⎡ ⎛ k 500ν ⎞ 4σ ω 2 k ⎤ ⎪⎫
4
⎫
⎪⎪ ⎪
F1 = tanh ⎨⎨min ⎢ max ⎜⎜ * , 2 ⎟⎟ , 2 ⎥⎬ ⎬ (15)
⎪⎪
⎩⎩
⎢⎣ ⎝ β ωy y ω ⎠ CDkω y ⎥⎦ ⎪⎭ ⎪⎭

In order to obtain the model constants of the k-ω SST, φ1 is considered as any constant in the
original model (Wilcox k-ω model) and φ2 as any constant in the transformed model (standard
k-ε model). Then, the constants are obtained via φ = φ1 F1 + φ2 (1 − F1 ) . The parameter F1 is
designed to be one in the near wall region which activates the original model and zero away
from the surface. The constants for the k-ω SST model are given in Table 3. The production
term is defined as:
⎛ ∂u ⎞
Pk = min ⎜ τij i ,10β* kω ⎟ (16)
⎜ ∂x ⎟
⎝ j ⎠

where τij is the turbulent shear stress term defined by Eq. (8). F2 is a function that is
respectively zero for free shear layer and one for boundary layer flows.

6
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

⎡⎡ ⎛ 2 k 500ν ⎞ ⎤ ⎤
2

F2 = tanh ⎢ max ⎜⎜ * , 2 ⎟⎟ ⎥ ⎥
⎢ (17)
⎢ ⎢⎣ ⎝ β ωy y ω ⎠ ⎥⎦ ⎥⎦
⎣

CD kω is the positive portion of the cross diffusion term.

⎛ 1 ∂k ∂ω ⎞
CD kω = max ⎜ 2ρσω 2 ,10 −10 ⎟ (18)
⎝ ω ∂x i ∂x i ⎠

All the computations have been carried out using the FLUENT software. The SIMPLE
algorithm was used for the calculation of the pressure field. Furthermore, the nonlinear
convective term in all transport equations was approximated using the MUSCL third-order
scheme. The first-order implicit scheme was employed for the discretization of the time
derivative terms.
Table 3: Empirical constants for the k-ω models
α1 α2 β1 β2 β* σk1 σk2 σω1 σω2

59 0.44 3 4 0.0828 0.09 0.85 1 0.5 0.856

To ensure the accuracy of the results, three different grid sizes were investigated using the k-
ω SST turbulence model, see Table 4. The grids in radial direction are clustered near the wall
but in the axial direction are uniformly distributed. As shown in Fig. 2, computations in all
three grids resulted in similar pressure variations. The second mesh with (40×1000) grid
nodes was selected for the turbulent flow simulations. For the laminar case, a computational
grid consisting of 3000×18 nodes in the axial and radial directions was selected.
Table 4: Number of grids
Case No. No. of Grids: (radial × axial)
1 20×500
2 40×1000
3 60×2000

To simulate water hammer phenomena due to the sudden valve closure, first steady state
computations were performed using the pressure inlet and the pressure outlet boundary
conditions. When the solution converged, the pressure outlet boundary condition was
switched to the wall boundary condition and the simulation continued with the unsteady
scheme. The time step of the transient solution was 10-4 sec.

3. 1 Laminar water hammer

The numerical results for the laminar water hammer are presented in Figs. 3 and 4. As clearly
seen in Fig. 3, the computed distribution of head at the mid-point between inlet and outlet of
the pipe is in excellent agreement with the experimental data of Holmboe et al. [1]. It can be
seen that at 0<t<0.7L/a, the pressure head does not change with time. This means that the
pressure wave still does not reach the mid-section of the pipe. At t=0.7L/a, the wave reflected
from the valve passes the midpoint and rises the head pressure. At t=1.7L/a, the pressure wave
reflected from the reservoir passes from the midpoint and thus reduces the head. By

7
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

increasing the time, at t=2.7L/a, the wave reflected from the valve passes the midpoint. In this
condition, the head reduces to minimum level. Finally, at t=4L/a, the second reflected wave
from the reservoir passes the midpoint which is an initial condition for the second wave cycle.
Fig. 4 shows the variation in the velocity profile at the mid-section at different times. The
results are in agreement with Wahba [21] numerical simulations, who simulated the same case
using the forth order Runge-Kutta scheme for the temporal discritization and the second order
central difference method for the spatial derivatives. It can be seen that at t=0 the velocity
profile is similar to the fully-developed velocity profile. By traveling the wave in a
reciprocating manner, the velocity magnitude changes with time. However, in all cases, away
from the wall, the velocity profile keeps its original shape. It should be mentioned that Wahba
[21] solved water hammer equations in their work while here we use the full Navier-Stokes
equations.

Fig. 2: The pressure head based on three grid numbers

Fig. 3: Head of the flow at the mid-section as a function of time, Experimental: Experimental results of Holmboe
et al. [1] , L: pipe length, a: speed velocity, V0: initial water velocity
The distribution of the wall shear stress for the laminar flow is shown in Fig. 5. The peaks in
the shear stress occur as the wave passes from the selected location. In between, the shear
stress reduces to a new steady condition.

8
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

Fig. 4: The change of velocity profile at mid section as a function of time, V0 initial water velocity, R pipe radius

Fig. 5: Shear stress variation on the wall, L: pipe length, a: speed velocity, V0: initial water velocity

3. 2 Turbulent water hammer

In Fig. 6, the time history of head is compared with the experimental data of Bergant et al. [2]
at mid-point between the inlet and outlet of the pipe. While results of both turbulence models
are more or less similar, the head predictions of the low-Re SST k-ω model are in better
agreement with the experimental data. Note that the second and third peaks in the head
distribution are somewhat overpredicted by the high-Re RNG k-ε model, but well captured by
the low-Re k-ω SST model.
In Fig. 7, the predicted velocity profiles at the mid-section of the pipe are compared with the
numerical results of Zhao et al. [22] who used the low-Re k-ε turbulence model of Fan et al.
[23] for simulations. As expected, the RNG k-ε model with wall functions is unable to capture
the variation of the velocity profile close to the wall. The main difference between the laminar
and turbulent cases lies in the near wall behavior of the flow. The strength of the reverse flow
is stronger in the turbulent flow. Moreover, the region of the reversed flow in the turbulent
flow is confined to a smaller region at the pipe wall.

9
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

Fig. 6: Head of the flow at the mid-section as a function of time, Experimental [2]: Experimental data of Bergant
et al., Numerical Simulation with RNG k-ω Turbulence model, Numerical simulation with k-ω
SST turbulence model, L: pipe length, a: wave speed , V0: initial water velocity
Fig. 8 shows the wall shear stress distribution at the mid-section of the pipe. It is seen that the
RNG k-ε model is unable to capture the wall shear stress variation correctly which is due to
the use of wall functions.

Fig. 7: The change of velocity profile at the mid-section as a function of time, Zhao et al. [22]: Numerical
simulation of Zhao et al., Numerical Simulation with RNG k-ε Turbulence model, Numerical simulation
with k-ω SST turbulence model. V0 initial water velocity, R pipe radius

4 CONCLUSIONS
It this paper, the full Navier-Stokes equations are used to simulate water hammer
phenomenon in the laminar and turbulent regimes. The numerical results for both test cases
are found to be in excellent agreement with the reported experimental data in the literature.
For the turbulent water hammer, the results of low-Re k-ω SST turbulence model are found to
be in better agreement with the experimental data. The velocity profiles tend to maintain their
initial shape except near the wall. The velocity profiles show a strong reverse flow near the
pipe wall which is stronger in the turbulent flow. Further 3D computations should be
performed to consider the effects of valve closure on the predictions.

10
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

Fig. 8: Shear stress variation on the wall, Zhao et al. [22]: Numerical simulation of Zhao et al., L: pipe length, a:
speed velocity, V0: initial water velocity

5 ACKNOWLEDGMENTS
The authors wish to thank University of Tehran (UT) and Luleå University of Technology (LUT)
for providing financial support of this work.

6 REFERENCES
[1] Holmboe, EL., Rouleau, WT., 1967, The effect of viscous shear on transients in liquid
lines, Journal of Basic Engineering, Transactions of the ASME, Vol. 89, No. 1, pp. 174–180.
[2] Bergant, A., Simpson, A. R., and Vitkovsky, J., 2001, Developments in Unsteady Pipe
Flow Friction Modelling, J. Hydraul. Res. Vol. 39, No. 3, pp. 249–257.
[3] Joukowski, N., 1904, Water Hammer, translated by Miss O. Simin, Proceedings of
American TFnier TForA's Assoc., Vol. 24, pp. 365 -368.
[4] Zielke, W., 1968, Frequency-dependent friction in transient pipe flow, J. Basic Eng., Vol.
90, No. 1, pp. 109–115.
[5] Brunone, B., Golia, U. M., and Greco, M., 1991, Some Remarks on the Momentum
Equation for Fast Transients, Proc. Int. Conf.on Hydr. Transients with Water Column
Separation, IAHR, Valencia, Spain, pp. 201–209.
[6] Safwat, H., 1972, On the Elastic Behavior of the Pipe Wall for Water Hammer
Applications, Nuclear Engineering and Design, pp. 85-94.
[7] Brunone, B., Karney, B. W., Mecarelli, M., Ferrante, M., 2000, Velocity Profiles and
Unsteady Pipe Friction in Transient Flow, J. of Water Resources Planning and Management,
pp. 236-244.
[8] Ghidaoui, M. S., and Kolyshkin, A. A., 2001, Stability Analysis of Velocity Profiles in
Water-Hammer Flows, Journal of Hydraulic Engineering, Vol. 127, No. 6, pp. 499–512.
[9] Streeter, V. L. and Wylie, E. B., 1985, Fluid Mechanics, 8th Edition, McGraw Hill New
York.
[10] Ghidaoui, M., Zhao, M., Mclnnis, D. A., Axworthy, D. H., 2005, A Review of Water
Hammer Theory and Practice, Journal of Applied Mechanics Review, Vol.58, No.1, pp.49-76.
[11] Pezzinga, G., 2000, Evaluation of unsteady flow resistances by quasi-2D or 1D models, J
Hydr. Engng., Vol. 126, No. 10, pp. 778–785.

11
 S. D. Saemi, M. Raisee, M. Cervantes and A. Nourbakhsh

[12] Trikha, A. K., 1975, An efficient method for simulating frequency-dependent friction in
transient liquid flow, Journal of Fluid Engineering, ASME, Vol. 97, No. 1, pp. 97–105.
[13] Vardy, A. E., and Brown, J. M. B., 2003, Transient Turbulent Friction in Smooth Pipe
Flows, J. Sound and Vibration, Vol. 259, No. 5, pp. 1011-1036.
[14] Vardy, A. E., and Brown, J. M. B., 2004, Transient Turbulent Friction in Rough Pipe
Flows, J. Sound and Vibration, Vol. 270, pp. 233-257.
[15] Cebeci, T., Smith, A. M. O., 1974, Analysis of turbulent boundary layers, Academic
Press, NewYork.
[16] Kita, Y., Adachi, Y., Hirose, K., 1980, Periodically oscillating turbulent flow in a pipe,
Bulletin JSME, Vol. 23, No. 179, pp. 656–664.
[17] Baldwin, B.S., Lomax, H., 1978, Thin layer approximation and algebraic model for
separated turbulent flows, AIAA Paper, pp. 78–257.
[18] Vardy, A. E., Hwang, K. L., 1991, A characteristic model of transient friction in pipes, J.
Hydr. Res., Vol. 29, No. 5, pp. 669–685.
[19] Silva-Araya, W. F., Chaudhry, M. H., 1997, Computation of Energy Dissipation in
Transient Flow, Journal of Hydraulic Engineering, American Society of Civil Engineers, Vol.
123, pp. 108–115.
[20] Zhao, M., Ghidaoui, M.S., 2003, Efficient quasi two-dimensional model for water
hammer problems, J. Hydr. Engng., Vol. 129, No. 12, pp. 1007–1013.
[21] Wahba, E. M., 2006, Runge-Kutta Time-Stepping Schemes with TVD Central
Differencing for the Water Hammer Equations, Int. J. Numerical Methods in Fluids, Vol. 52,
pp. 571-590.
[22] Zhao, M., Ghidaoui, M. S., 2006, Investigation of Turbulence Behavior in Pipe Transient
Using a k-ε Model, J. of Hydraulic Research, Vol. 44, No. 5, PP. 682-692.
[23] Fan, S., Lakshminarayana, B., and Barnett, M., 1993, Low-Reynolds-Number Model for
Unsteady Turbulent Boundary-Layer Flows, AIAA J., Vol. 31, No. 10, pp. 1777–1784.
[24] Riasi, A., Nourbakhsh, A., Raisee, M., 2009, Unsteady turbulent pipe flow due to water
hammer using k-ω turbulence model, J. of Fluids Engineering, Vol. 47, No. 4, pp. 429-437.
[25] Wylie, E. B., Streeter, V. L., and Lisheng, S., 1993, Fluid Transient in Systems, Prentice-
Hall, Englewood Cliffs.
[26] Chaudhry, M. H., 1987, Applied Hydraulic Transients, Van Nostrand Reinhold, N Y.
[27] Tijsseling, A. S, Lambert, M. F., Simpson, A. R., Stephens M. L., Vítkovský, J. P.,
Bergant, A., 2008, Skalak’s extended theory of water hammer, Journal of Sound and
Vibration, Vol. 310, No. 3, pp. 718-728.
[28] Chaudhry, M. H., and Hussaini, M. Y., 1985, Second-order accurate explicit finite-
difference schemes for water hammer analysis, J. Fluids Eng., Vol. 107, pp. 523–529.
[29] Guinot, V., 2002, Riemann Solvers for Water Hammer Simulations by Godunov Method,
Int. J. Numer. Methods Eng., Vol. 49, pp. 851–870.
[30] Zhao, M., Ghidaoui, M., 2004, Godunov-Type Solutions for Water Hammer Flows,
Journal of Hydraulic Engineering, Vol. 130, pp. 341-348.
[31] Yakhot, V., Orszag, S.A., Thangam, S., Gatski, T.B. & Speziale, C.G., 1992,
Development of turbulence models for shear flows by a double expansion technique, Physics
of Fluids A, Vol. 4, No. 7, pp1510-1520.
[32] Menter, F. R., 1994, Two-Equation Eddy-Viscosity Turbulence Models for Engineering
Applications, AIAA Journal, Vol. 32, No 8. pp. 1598-1605.

12