Sunteți pe pagina 1din 12

Acta Mech 216, 75–86 (2011)

DOI 10.1007/s00707-010-0358-x

E. M. Wahba

A computational study of viscous dissipation and entropy


generation in unsteady pipe flow

Received: 3 February 2010 / Revised: 20 April 2010 / Published online: 29 June 2010
© Springer-Verlag 2010

Abstract Numerical simulations of transient entropy generation in a reservoir-pipe-valve system are presented.
The flow transient is initiated through sudden closure of the downstream valve. An unsteady two-dimensional
water hammer model is adopted. Time integration is performed using the classical fourth-order Runge–Kutta
method while the spatial terms are discretized using central difference expressions. Entropy generation is
shown to depend on a non-dimensional parameter representing the ratio of the viscous diffusion time scale to
the pipe period. For small values of the non-dimensional parameter, entropy generation is rapidly attenuated
from its steady-state value to zero while for large values, entropy generation persists for a much longer time.
Moreover, for large values of the non-dimensional parameter, excessive entropy generation rates are realized
during the transient which are several orders of magnitude higher than the steady-state rate. Such a behavior
is attributed to elevated transient shear stress values in the near wall region which result in excessive viscous
dissipation and hence higher entropy generation rates. Finally, it is shown that during the transient, the location
of maximum entropy generation is no longer restricted to the pipe wall.

List of symbols
a Wave speed
D Pipe diameter
g Gravitational acceleration
H Pressure head
H0 Pressure head at steady state
L Pipe length
r Radial location across the pipe
R Pipe radius
Re Reynolds number = V0 D/ν
S Instantaneous entropy generation rate per unit volume
Sav,ss Average entropy generation rate per unit volume at steady state
t Time
tDiff Viscous diffusion time scale (D 2 /ν)
tp Pipe period (L/a)
T Temperature

E. M. Wahba (B)
Mechanical Engineering Department, American University of Sharjah, Sharjah 26666, UAE
E-mail: emwahba@yahoo.com
76 E. M. Wahba

u Axial component of velocity


V Average cross-sectional velocity
V0 Average cross-sectional velocity at steady state
x Axial location along the pipe
v Radial component of velocity
β Non-dimensional parameter = (a D 2 )/(Lν)
ϕ Viscous dissipation function
μ Dynamic viscosity
ν Kinematic viscosity (μ/ρ)
ρ Fluid density
τ Stress tensor

1 Introduction

Steady laminar flow in circular pipes can be solved analytically using Navier–Stokes equations. The problem
is well known in the literature as the Hagen-Poiseuille flow [1]. The analytical solution of the velocity profile
is parabolic with zero velocity at the pipe wall due to the no-slip condition. The average velocity in the pipe
is half the value of the maximum velocity attained at the pipe axis. With the velocity profile known, analytical
expressions for viscous dissipation and entropy generation in the pipe are readily obtained. Entropy generation
profiles are parabolic, with a maximum value at the pipe wall and zero value at the axis [2]. Hence, under
laminar flow conditions, the entire flow field participates in the production of entropy.
Moreover, entropy generation was analyzed for flow in ducts subjected to either constant heat flux [3] or
constant wall temperature [4]. The effect of variable viscosity on entropy generation was presented by Sahin
[5], while irreversibilities due to non-Newtonian flow through pipes were studied by Mahmud and Fraser [6].
Entropy generation in the developing region, i.e. in the pipe entrance length, was also analyzed in [7] using a
two-dimensional Navier–Stokes solver. In all these studies, the flow in the pipe was assumed to be steady.
However, it is entirely normal to experience changes in flow conditions within a piping system. Such
changes could be caused by the closure of a valve or the shutdown of a pump. In such cases, the flow in the
pipe is no longer steady. For gradual changes in flow conditions, rigid water column theory [8] provides fairly
accurate results. However, for rapid changes in flow conditions, the results obtained from rigid water column
theory are no longer accurate, and the elastic theory [9] needs to be applied, which takes into consideration
the elasticity of the pipe and the compressibility of the liquid.
Using elastic theory, fluid transients are frequently simulated using one-dimensional (1-D) models [8,9]
in which the pipe average velocity is considered to be a function of the axial location only. The (1-D) model
provides an accurate prediction of the magnitude and shape of the first pressure peak. However, it underes-
timates the attenuation of the subsequent pressure peaks. This is due to the inadequate representation of the
viscous term in the (1-D) formulation, because (1-D) water hammer models assume that the velocity profile
in unsteady flow would be similar to that in steady flow. This is certainly not the case for fast transients,
in which strong adverse pressure gradients could result in large regions of flow separation and reversal in
the pipe. Experimental evidence of the complexity of unsteady velocity profiles was reported by Brunone
et al. [10].
A significant improvement on the modeling of unsteady friction in (1-D) water hammer models was intro-
duced by Zielke [11], where the instantaneous wall shear stress was shown to be equal to the quasi-steady
value plus a term consisting of weighted past velocity changes. To provide further insight into the behavior of
unsteady friction during fluid transients, a number of two-dimensional numerical simulations for unsteady flow
in elastic pipes were presented in the literature [12–18]. The (2-D) simulations reveal that the velocity profile
under transient conditions is no longer similar to the steady-state velocity profile [12]. Moreover, through
order of magnitude and dimensional analysis of the (2-D) model, Wahba [17] showed that viscous effects in
fluid transients are not controlled by the Reynolds number but by another non-dimensional parameter which
represents the ratio of the viscous diffusion time scale to the pipe period.
Entropy generation and the second law of thermodynamics provide an alternative approach for studying
frictional effects in pipes because of the inherent irreversibility associated with the process [2]. Using extended
irreversible thermodynamics and a second law approach, various model equations were developed for unsteady
friction [19,20]. A common theme between these models is their need for calibration, i.e. coefficients are pres-
ent in the model equation and suitable values are sought for them, usually through a trial-and-error procedure.
A computational study of viscous dissipation and entropy generation in unsteady pipe flow 77

A better understanding of transient entropy generation during the water hammer event would certainly help in
providing further insight into these entropy-based models.
In the present study, two-dimensional water hammer simulations will be used to evaluate transient viscous
dissipation and entropy generation due to sudden valve closure in a reservoir-pipe-valve system. To achieve
this, the governing equations for the (2-D) water hammer model are stated first, and the corresponding assump-
tions are underlined. The numerical scheme is then introduced and used to simulate the reservoir-pipe-valve
system. Moreover, a numerical parametric study is conducted to determine the effect of the non-dimensional
parameter, introduced by Wahba [17], on transient entropy generation rates.

2 Governing equations and numerical methods

In the present study, the quasi two-dimensional model of Pezzinga [14] is adopted to model unsteady pipe
flow. In this model, the flow is assumed to be axisymmetric, and the pressure is assumed to be independent of
the radial direction. Under such assumptions, the mass and momentum conservation laws take the following
form:
∂H ∂H a2 ∂ V
+V + = 0, (1)
∂t ∂x g ∂x
∂u ∂u ∂H 1 ∂ (r τr x )
+u = −g + . (2)
∂t ∂x ∂x ρr ∂r
The elasticity of the pipe and the compressibility of the liquid is included through the wave speed, a, which
appears in the mass conservation Eq. 1. Note that in the above equations, the flow inside the pipe is assumed to
be one-directional; hence, the radial component of velocity, v, is zero. For a Newtonian fluid, the shear stress
term (τr x ) is related to the strain rate as follows:
∂u
τr x = ρν . (3)
∂r
The average velocity (V) is evaluated from the local velocity (u) at each cross-section using the formula
R
u2πr dr
V = 0 . (4)
π R2
Once the velocity profile is obtained, the viscous dissipation function and the corresponding entropy generation
rate could be evaluated from the following expressions:
 2  2
∂u ∂u
=2 + , (5)
∂x ∂r
μ
S = . (6)
T
A more instructive form of the governing equations could be obtained by re-writing them using the following
non-dimensional variables:
r t x ⎫
r∗ = , t∗ = , x∗ = , ⎪

D L/a L
u V H SD T ⎪2 . (7)
u∗ = ,V∗ = , H∗ = , S∗ = ⎭
V0 V0 aV0 /g μU 2
Combining Eqs. (5) and (6) and applying the non-dimensional variables to the combined equation and to
Eqs. (1) and (2) results in the following set of non-dimensional governing equations:
 
∂ H∗ V0 ∗∂H
∗ ∂V ∗
+ V + = 0, (8)
∂t ∗ a ∂x∗ ∂x∗

    ∂ r ∗ ∂u∗∗
∂u ∗ V0 ∂u ∗ ∂ H ∗ Lν 1 ∂r

+ u∗ ∗ = − ∗ + ∗
, (9)
∂t a ∂x ∂x aD 2 r ∂r ∗
 2  ∗ 2  ∗ 2
D ∂u ∂u
S∗ = 2 + . (10)
L ∂x∗ ∂r ∗
78 E. M. Wahba

2
Present Method
Experiment
1.5

(H−H )/(aV /g)


0.5

o
o 0

−0.5

−1

−1.5

−2
0 5 10 15
at/L
Fig. 1 Pressure–time history at the valve (β = 576.5)

From Eq. (8), the non-dimensional parameter aLν D


, termed β hereafter, emerges as an important
non-dimensional parameter that would have significant effect on viscous dissipation and hence entropy gener-
ation in unsteady pipe flow. The term β can be physically interpreted in terms of the various time scales present
in the fluid transient. Strictly speaking, two time scales exist, the viscous diffusion time scale tDiff = D 2 /ν
and the pipe period t P = L/a. Hence, β represents the ratio of the viscous diffusion time scale to the pipe
period.
The numerical scheme developed by Wahba [12] is used to solve the non-dimensional (2-D) water hammer
model. The mass and axial momentum equations are integrated in time using the classical fourth-order Runge–
Kutta method. Second-order accurate finite difference expressions are used to evaluate the spatial derivatives.
For numerical stability purposes, second-order dissipative terms are added in regions of high gradients. The
dissipative terms are switched off in regions of low gradients using a numerical switch. A detailed explanation
of the numerical procedure is given in [12].

3 Model validation

The present numerical scheme is first validated against the experimental results of Holmboe and Rouleau [21].
The physical system consists of a copper pipe connecting an upstream constant head reservoir to a downstream
fast acting valve. The pipe has an inner diameter of 0.025 m and a length of 36.09 m. Pressure signals are
experimentally recorded directly upstream of the valve and at the pipe midpoint. The operating fluid is high
viscosity oil (ν = 39.77 × 10−6 m2 /s), and the wave speed is measured to be 1,324 m/s. The flow inside the
pipe is laminar with a Reynolds number value of 82 and a β value of 576.5.
The results of the present numerical simulation agree well with the experimental results of Holmboe and
Rouleau [21], as evident from Figs. 1 and 2. In these two figures, the numerical and experimental transient
pressure signals are compared at the valve and at the pipe midpoint. Moreover, present numerical results for
the instantaneous wall shear stress at the pipe midpoint are in good agreement with previous numerical results
of Vardy and Hwang [15], as can be seen in Fig. 3. Interestingly, the wall shear stress is shown to attain values
that are nearly ten times the value of the steady-state shear stress.
However, it should be noted that the experimental results of Holmboe and Rouleau [21] are recorded only
over a limited number of pipe periods. Hence, to validate the effectiveness of the present method in modeling
the attenuation of fluid transients over extended periods of time, the experimental results of Brunone et al. [22]
are considered. The experimental setup consists of a copper pipe of 20 mm diameter and 141.07 m length. An
air vessel is used as a constant head supply reservoir at the upstream side, with the transient being generated
by the closure of a ball valve placed at the downstream end and discharging into a free surface tank. The wave
A computational study of viscous dissipation and entropy generation in unsteady pipe flow 79

2
Present Method
Experiment
1.5

(H−H )/(aV /g)


0.5

o
o 0

−0.5

−1

−1.5

−2
0 5 10 15
at/L
Fig. 2 Pressure–time history at the midpoint (β = 576.5)

8
Present Method
6 Vardy and Hwang (1991)

0
ss
w w

−2
τ /τ

−4

−6

−8

−10

−12
0 2 4 6 8 10
at/L
Fig. 3 Wall shear stress–time history at the midpoint (β = 576.5)

speed is 1,120 m/s, and the initial Reynolds number is 815. The value of the non-dimensional parameter β for
this experimental test case is 3,176. The downstream ball valve is closed in 0.11 s and the pressure is recorded
at a point located 42 m from the upstream end. Numerical simulations for this test case are reported in Fig. 4
and compared with the experimental results. As evident from Fig. 4, the present method effectively captures
the long-term attenuation of the transient.

4 Entropy generation rates and the effect of the non-dimensional parameter (β)

Radial profiles of entropy generation at the pipe midpoint for the Holmboe and Rouleau [21] test case are
given in Fig. 5 at several time instants. Figure 5 shows elevated entropy generation rates at the pipe wall,
while decreasing to zero at the pipe axis. Note that in all figures of the present study, entropy generation rates
are normalized by the average value of the steady-state entropy generation. The temporal variation of entropy
80 E. M. Wahba

1
Present Method
0.8 Experiment

0.6

0.4

(H−H )/(aV /g)


0.2

o
o 0

−0.2

−0.4

−0.6

−0.8

−1
0 20 40 60 80
at/L
Fig. 4 Long-term attenuation of the pressure transient at the valve (β = 3,176)

0.5

0.45

0.4

0.35

0.3
r/D

0.25

0.2

0.15
t=0
0.1 t=L/a
t=2L/a
0.05 t=3L/a
t=4L/a
0
0 5 10 15 20 25 30
S/Sav,ss

Fig. 5 Transient entropy generation profiles at the midpoint (β = 576.5)

generation at the pipe midpoint for r = R and r = R/2 is given in Figs. 6 and 7 respectively. Note that at
certain instants of time, entropy generation rates at the pipe wall are nearly two orders of magnitude higher
than the steady-state entropy generation. At r = R/2, transient entropy generation rates remain lower than
their steady-state value.
Better explanation of such behavior could be obtained from the transient shear stress profiles at the pipe
midpoint, given in Fig. 8. It is clear from this figure that the elevated values of entropy generation rates at the
pipe wall are due to excessive wall shear stress values. It is interesting to note that maximum shear stress and
maximum entropy generation are not always located at the pipe wall. As can be seen from Figs. 5 and 8, at
time t = 3L/a, maximum shear stress and maximum entropy generation are located at r = 0.85R.
For the Holmboe and Rouleau [21] laminar flow case, the value of the non-dimensional parameter is
(β = 576.5). In the following, a parametric study is performed to determine the effect of varying β on the
transient entropy generation rates. Two new cases are simulated which have the same Reynolds number as the
A computational study of viscous dissipation and entropy generation in unsteady pipe flow 81

200

180

160

140

120

S / Sav,ss
100

80

60

40

20

0
0 5 10 15
at/L
Fig. 6 Entropy generation–time history at the midpoint (r = R, β = 576.5)

0.5

0.45

0.4

0.35
av,ss

0.3
S/S

0.25

0.2

0.15

0.1

0.05

0
0 5 10 15
at/L
Fig. 7 Entropy generation–time history at the midpoint (r = R/2, β = 576.5)

0.5

0.45

0.4

0.35

0.3
r/D

0.25

0.2

0.15
t=0
0.1 t=L/a
t=2L/a
0.05 t=3L/a
t=4L/a
0
−0.3 −0.2 −0.1 0 0.1 0.2 0.3 0.4
2
τ / (ρ V )
Fig. 8 Transient shear stress profiles at the midpoint (β = 576.5)
82 E. M. Wahba

Table 1 Summary of simulated cases

Case number β = (a D 2 )/(Lν) Re = (V0 D)/(ν)


1 576.5 82
2 57.65 82
3 5,765 82

β=57.65
av,ss

5
S /S
w

0
0 5 10 15
at/L
200
av,ss

β=576.5
S /S

100
w

0
0 5 10 15
at/L
av,ss

1000 β=5765
S /S

500
w

0
0 5 10 15
at/L
Fig. 9 Entropy generation–time history at the midpoint for different β values (r = R)

0.5
av,ss

β=57.65
/S
mid
S

0
0 5 10 15
at/L
av,ss

0.5 β=576.5
/S
mid
S

0
0 5 10 15
at/L
av,ss

0.5
/S
mid

β=5765
S

0
0 5 10 15
at/L
Fig. 10 Entropy generation–time history at the midpoint for different β values (r = R/2)

previous case (Re = 82), but in which the value of the parameter β is varied by varying the pipe length. As
a summary, all three simulated cases are given in Table 1. The results for cases (2) and (3) and comparisons
with case (1) are given in Figs. 9 and 10 in terms of the temporal variation of entropy generation at the pipe
midpoint for r = R and r = R/2.
For case (2), the value of β is lowered to 57.65. As can be seen from Figs. 9 and 10, entropy generation
rates are rapidly attenuated, both at r = R/2 and at the pipe wall r = R, compared to the higher β cases of
A computational study of viscous dissipation and entropy generation in unsteady pipe flow 83

0.5

0.45

0.4

0.35

0.3
r/D
0.25

0.2

0.15
t=0
0.1 t=L/a
t=2L/a
0.05 t=3L/a
t=4L/a
0
0 0.5 1 1.5 2
S/S
av,ss

Fig. 11 Transient entropy generation profiles at the midpoint (β = 57.65)

0.5

0.45

0.4

0.35

0.3
r/D

0.25

0.2

0.15
t=0
0.1 t=L/a
t=2L/a
0.05 t=3L/a
t=4L/a
0
−0.1 −0.05 0 0.05 0.1
2
τ/(ρV )
Fig. 12 Transient shear stress profiles at the midpoint (β = 57.65)

576.5 and 5,765. Moreover, transient entropy generation and shear stress profiles at the pipe midpoint are given
in Figs. 11 and 12 respectively. These figures show reduced levels of shear stress and entropy generation when
compared to the higher β cases. At time t = 3L/a, it is again clear that the location of maximum shear and
entropy generation is no longer at the pipe wall, instead it is located at r = 0.65R compared to (r = 0.85R)
for case (1).
For case (3), the value of β is set to 5,765, a value much larger than cases (1) and (2). Figure 9 shows that
transient entropy generation rates at the pipe wall, at certain time instants, are nearly three orders of magnitude
higher than the steady-state entropy generation rate. Moreover, Figs. 9 and 10 show that the attenuation of the
transient entropy generation rate is insignificant even over extended time periods, both at r = R/2 and at the
pipe wall r = R. Figures 13 and 14 show entropy generation and shear stress profiles at different time instants.
Similar to cases (1) and (2), at time t = 3L/a, the maximum shear stress and entropy generation is not located
84 E. M. Wahba

0.5

0.45

0.4

0.35

0.3

r/D 0.25

0.2

0.15
t=0
0.1 t=L/a
t=2L/a
0.05 t=3L/a
t=4L/a
0
0 50 100 150 200 250 300 350
S/Sav,ss

Fig. 13 Transient entropy generation profiles at the midpoint (β = 5,765)

0.5

0.45

0.4

0.35

0.3
r/D

0.25

0.2

0.15
t=0
0.1 t=L/a
t=2L/a
0.05 t=3L/a
t=4L/a
0
−1 −0.5 0 0.5 1 1.5
2
τ / (ρ V )
Fig. 14 Transient shear stress profiles at the midpoint (β = 5,765)

at the pipe wall, instead it is located at r = 0.95R. It is clear from Figs. 13 and 14 that up till t = 4L/a, the
shear stress and entropy generation in the bulk core of the flow are virtually unaffected. Only the near wall
region is affected by the fluid transient, resulting in extremely high shear stress and entropy generation rates,
when compared to cases (1) and (2).
This behavior could be explained through the definition of β as the ratio of viscous diffusion time scale
to the pipe period. For large values of β, as in case (3), the viscous diffusion time scale is much larger than
the pipe period, and hence it will take extended periods of time for viscous effects to diffuse through the pipe
cross-section as can be seen from the shear stress profiles in Fig. 14. On the other hand, for small values of β,
as in case (2), the viscous diffusion time scale is comparable to the pipe period, and hence viscous effects are
rapidly diffused through the pipe cross-section as can be seen from the shear stress profiles in Fig. 12.
The significant effect of the value of β on instantaneous shear stress profiles for laminar fluid transients
draws an interesting comparison with another non-dimensional parameter, P, for turbulent fluid transients that
A computational study of viscous dissipation and entropy generation in unsteady pipe flow 85

2−D Quasi−steady 1−D β=57.65

(H−Ho)/(aVo/g)
1.5
1
0.5
0
0 5 10 15
at/L
β=576.5
(H−Ho)/(aVo/g)

1
0
−1
0 5 10 15
at/L
β=5765
(H−Ho)/(aVo/g)

1
0
−1
0 5 10 15
at/L
Fig. 15 Pressure–time history at the valve for different β values

was proposed by Ghidaoui et al. [23]. In their paper, Ghidaoui et al. [23] proposed the following parameter P
for assessing the accuracy of quasi-steady turbulence modeling in water hammer problems:

2D/ f V0
P= .
L/a
The parameter P represents the ratio of the turbulent diffusion time scale to the wave propagation time scale.
Ghidaoui et al. [23] concluded that, for large values of P, the structure and intensity of turbulence during the
transient remains very similar to that of the steady-state flow, and the adoption of steady-state-based turbu-
lence models is justified. A similar conclusion could be drawn for laminar fluid transients, where for large
values of β, the simple quasi-steady approximation could be used to model the unsteady friction term without
significant loss of accuracy. This conclusion is validated through Fig. 15, which shows a comparison between
the time histories of the (1-D) and (2-D) computed pressure heads at the valve for different values of β.
For the (1-D) model, the classical method of characteristics is used [9], and the viscous term is evaluated using
the quasi-steady approximation. Figure 15 clearly shows that for large values of β, the quasi-steady approach
provides a satisfactory approximation for unsteady friction. Finally, it is interesting to show that the parameter
β for laminar fluid transients could be re-cast in a form similar to that of P, by noting that for laminar flows
f ∼1/Re and hence
D/ f V0
β= .
L/a

5 Conclusions

Entropy generation rates in unsteady pipe flows initiated by sudden closure of a downstream valve are numer-
ically simulated in the present study. It is found that entropy generation depends on a non-dimensional
parameter β, representing the ratio of the viscous diffusion time scale to the pipe period. For small val-
ues of β, transient entropy generation rates are comparable to the steady-state entropy generation and are
rapidly attenuated during the transient. However, for large β values, elevated transient entropy generation rates
are realized, which are nearly three orders of magnitude higher than the steady-state entropy generation. More-
over, through monitoring the transient entropy generation and shear stress profiles, it is shown that maximum
86 E. M. Wahba

entropy generation and shear stress values are not always located at the pipe wall but can be located within the
flow.

References

1. White, F.M.: Viscous fluid flow, 3rd edn. McGraw-Hill, New York (2005)
2. Bejan, A.: Entropy generation minimization, 1st edn. CRC press, Boca Raton (1995)
3. Nag, P.K., Kumar, N.: Second law optimization of convective heat transfer though a duct with constant heat flux. Int. J.
Energ. Res. 13, 537–543 (1989)
4. Sahin, A.Z.: Second law analysis of laminar viscous flow through a duct subjected to constant wall temperature. ASME J.
Heat Trans. 120, 76–83 (1998)
5. Sahin, A.Z.: The effect of variable viscosity on the entropy generation and pumping power in a laminar fluid flow through
a duct subjected to constant heat flux. Heat Mass Trans. 35, 499–506 (1999)
6. Mahmud, S., Fraser, R.A.: Inherent irreversibility of channel and pipe flows for non-Newtonian fluids. Int. Commun. Heat
Mass Trans. 29, 577–587 (2002)
7. Sahin, A.Z., Ben-Mansour, R.: Entropy generation in laminar fluid flow through a circular pipe. Entropy 5, 404–416 (2003)
8. Watters, G. Z.: Analysis and control of unsteady flow in pipelines. Butterworth, Stoneham (1984)
9. Wylie, E.B., Streeter, V.L.: Fluid Transients in Systems. Prentice Hall, New Jersey (1993)
10. Brunone, B., Karney, B.W., Mecarelli, M., Ferrante, M.: Velocity profiles and unsteady pipe friction in transient flow.
J. Water Resour. Plann. Manage. 126, 236–244 (2000)
11. Zielke, W.: Frequency-dependent friction in transient pipe flow. J. Basic Eng., Trans. ASME 90(1), 109–115 (1968)
12. Wahba, E.M.: Runge–Kutta time-stepping schemes with TVD central differencing for the water hammer equations. Int.
J. Numer. Methods Fluids 52(5), 571–590 (2006)
13. Brunelli, M.C.P.: Two-dimensional pipe model for laminar flow. J. Fluids Eng., ASME 127, 431–437 (2005)
14. Pezzinga, G.: Quasi-2D model for unsteady flow in pipe networks. J. Hydraulic Eng. 125, 676–685 (1999)
15. Vardy, A.E., Hwang, K.L.: A characteristics model of transient friction in pipes. J. Hydraulic Res. 29(5), 669–684 (1991)
16. Nathan, G.K., Tan, J.K., Ng, K.C.: Two-dimensional analysis of pressure transients in pipelines. Int. J. Numer. Methods
Fluids 8(3), 339–349 (1988)
17. Wahba, E.M.: Modelling the attenuation of laminar fluid transients in piping systems. Appl. Math. Modell. 32, 2863–2871
(2008)
18. Wahba, E.M.: Turbulence modeling for two dimensional water hammer simulations in the low Reynolds number range.
Comput. Fluids 38, 1763–1770 (2009)
19. Axworthy, D.H., Ghidaoui, M.S., McInnis, D.A.: Extended thermodynamics derivation of energy dissipation in unsteady
pipe flow. J. Hydraulic Eng. 126, 276–287 (2000)
20. Kucienska, B., Seynhaeve, J., Giot, M.: Friction relaxation model for fast transient flows application to water hammer in
two-phase flow—the WAHA code. Int. J. Multiphase Flow 34, 188–205 (2008)
21. Holmboe, E.L., Rouleau, W.T.: The effect of viscous shear on transients in liquid lines. J. Basic Eng., Trans. ASME 89(1),
174–180 (1967)
22. Brunone, B., Ferrante, M., Cacciamani, M.: Decay of pressure and energy dissipation in laminar transient flow. J. Fluids
Eng., ASME 126, 928–934 (2004)
23. Ghidaoui, M.S., Mansour, S.G.S., Zhao, M.: Applicability of quasisteady and axisymmetric turbulence models in water
hammer. J. Hydraulic Eng., ASCE 128, 917–924 (2002)

S-ar putea să vă placă și