Beruflich Dokumente
Kultur Dokumente
ABSTRACT
In pumping installations, fluid transient computations are necessary to achieve safety, efficiency and
economy in design and operation. In some systems, where air content and air entrainment exist, such
computations become highly inaccurate when constant wave speed is assumed. In this paper, a numerical
model and a computational procedure have been developed to investigate the effects of air entrainment on
the pressure transient in pumping systems. Free gas in the fluid and cavitation at the fluid vapour pressure
were modeled in the form of variable wave speed model, which was numerically solved by the method of
characteristics. This model was tested for the case of pump trips due to power failures. The pressure
transient results obtained by this variable wave speed model were analyzed and compared with those results
obtained by constant wave speed model and with the experimental results of other investigators.
NOMENCLATURE
a wave speed
R C+ line intercept on x-axis
A1, A2, A3 constants for pump H-Q curve
S C- line intercept on x-axis
B1, B2, B3 constants for pump T-Q curve
T pump torque
C1, C2, C3 constants for pump η-Q curve t time
cl parameter describing pipe constraint V flow velocity
D mean diameter of pipe x distance along pipeline
E modulus of elasticity
α pipeline inclination
e local pipe wall thickness
η pump efficiency
ƒ friction factor
∆tk time step at kth time level
g gravitational acceleration
H gauge piezometric pressure head ∆x node point distance along pipeline
I pump set moment of inertia ε fraction of gas in liquid
k time level ε0 initial air void fraction
K bulk modulus of elasticity εg fraction of dissolved gas in liquid
Nik pump speed in rpm ρ density of fluid
N total number of node points Subscripts
np number of pumps in a pumping station
P pressure inside the pipe d dissolved
Q fluid flow rate e equivalent
r released
0 initial
1. INTRODUCTION
In pumping installation, fluid transient computation of hydraulic components. In these systems, air
is necessary to predict excessive transient pressures content and air entrainment always exist and affect
which may cause collapse of pipelines, and damage the pressure transient. Air in pumping systems
comes from three primary sources. The first source
T.S. Lee et al. / JAFM , Vol. 1, No. 1, pp. 55-61, 2008.
of air is trapped air pockets at the top of the pipe the effects of free gas in the fluid and cavitation at
cross-section at high points along pipe profile due the fluid vapour pressure on the pressure transient
to the incomplete removal of air or the progressive in the pipeline. This model is solved numerically
upward migration of air pockets. The second source by using the method of characteristics.
of air is free gas, dissolved gas in the flow which
contains approximately 2% volume of water. The
third source of air comes from mechanical
equipment in systems. This air may be forced into 2. VARIABLE WAVE SPEED MODEL
the systems as a result of falling jets of sewage into In this paper, the variable wave speed model
the pump sump, attached vortex formation, and assumes the existence of free entrained air content
adverse flow path towards operating pump. This air ε0 and dissolved gas content εg in the liquid at
may also be admitted though packing, air valves, atmospheric pressure. Some assumptions were
air vessel, etc. under vacuum conditions. applied in this model: (i) the gas-liquid mixture is
The effects of air entrainment on the pressure homogeneous, (ii) free gas bubbles in the liquid
transient in pumping systems were firstly studied follow a polytrophic compression law with n = 1.2-
by Whiteman and Pearsall (1959, 1962) in their 1.3, and (iii) the pressure within air bubbles is in
pump shut-down tests. In practice, analysis of equilibrium with the local fluid pressure. When the
pressure transient with air entrainment face many local pressure falls below the liquid vapour
difficulties owing to lack of information such as: pressure, an amount of dissolved gas content αrεg is
the location and size of air pockets in the pipelines; assumed to release instantaneously. The local
the amount of free air bubbles distributed in the pressure is kept constant and is equal to the vapour
liquid, and the physical process of release and pressure. When the local pressure recovers to a
absorption of air in the liquid. Numerical modelling value higher than the vapour pressure, an amount
of fluid transients with air entrainment has been the of gas content αdεg is assumed redissolved into the
subject of much research. The concentrate liquid.
vaporous cavity model (Brown 1968, Provoost The effective bulk modulus KT of the gas-liquid
1976) and the air release model (Fox 1972, Wylie mixture, including the pipe distensibility effect and
1980) have been proposed in literature showing pipe constraint condition cl is given by (Fox 1984,
reasonable prediction of pressure transient Lee 1991):
behaviours in pipeline systems. The variable wave
speed model used by Lee (1991), Borga et al 1 1 ε cl D
(2004) can improve the simulation of experimental = + + (1)
observations in terms of shape of the pressure KT K np eE
peaks, the frequency of the oscillations and the rate
of decay. When air is entrained such that the gas Along the pipeline, the fraction of air content
void fraction is significant and two phase motion depends on the local pressure and local air volume.
occurs, it become necessary to introduce multi- Therefore, in variable wave speed model, wave
phase modeling (Huygens et al, 1998, Fujii and speed is not constant and is calculated for each
Akagawa, 2000 and Lee et al, 2004). point i which has local pressure pi and air fraction
This paper aims to investigate the effects of air content εi:
entrainment on pressure transient in a typical
−1
⎡ ⎛ 1 ε ik cl D ⎞⎤
( )
pumping system. The system consists of a lower 2
a = ⎢ ρ w 1 − ε ik
k
⎜⎜ + k + ⎟⎥ (2)
eE ⎟⎠⎦
reservoir, a group of three pumps in parallel which i
has a check valve in each branch, and a pipeline ⎣ ⎝ K npi
system discharging into an upper reservoir. The
most dangerous case of pressure transient is the The initial free air fraction ε0 and dissolved gas
stop of all three pumps in the station due to a power fraction εg must be specified. Then, the initial
failure. In this case, the following events take variable wave speed along the pipeline is computed
place: (i) the flow rapidly diminishes to zero and through the absolute pressure distribution. This
then reverses, (ii) when the flow reverses, positive value is then used to calculate the pressure
pressure waves propagate downstream of the pump distribution for the next time level. The air fraction
towards the reservoir, and negative pressure waves for the next time level is given by (Lee 1991):
propagate upstream of the pipe towards the suction
of the pump, (iii) the pump rapidly losses its 1/ n 1/ n
⎛ pik ⎞ ⎛ p ⎞
forward rotation, and reverses, (iv) to prevent ε Tk +1 = ⎜⎜ ⎟ ε ik and ε 0k +1 = ⎜⎜ k0+1 ⎟⎟ ε 0
k +1 ⎟
(3a)
reverse flow through the pump, when the flow p
⎝ i ⎠ ⎝ pi ⎠
reverses the check valve is activated and closed. A
large pressure transient occurs in the pipeline. A
For pik +1 ≥ p g and ε Tk +1 ≤ ε 0k +1 + α r ε g :
variable wave speed model is proposed to include
56
T.S. Lee et al. / JAFM , Vol. 1, No. 1, pp. 55-61, 2008.
dx
=V ± a (6, 7) η ek +1 = C1 + (C 2 / n p )(Q0k +1 / N k +1 )
dt
( )(
+ C3 / n 2p Q0k +1 / N k +1 )
2
(15)
Then, the characteristic equations can be
approximated by finite difference expressions: Te = − I e dω / dt (16)
k +1 k +1
g H i − H R Vi − VR
+ where Hek+1 = H0k+1, Ie = npI, ω = 2πN, Q is the
aR ∆t k ∆t k flow rate, np is the number of pumps, A1, A2, A3, B1,
g f l VR | VR | B2, B3, and C1, C2, C3 are single pump constants.
+ VR sin α i + R =0 (8) He, Te, and ηe are the equivalent pump variables.
aR 2D
At time tk = k ∆t, Nk, Qk, Tk, Hk, and ηk are assumed
known. For time tk+1 = (k+1)∆t the computation
xi − xR
= VR + aR (9) for pump run down characteristic is given by the
∆t k following procedure:
k +1 k +1 1) Estimated a new pump speed:
g H i − H S Vi − VS
− +
aS ∆t k ∆t k 60∆t
N 0k +1 = N k − T k (17)
g f l VS | V S | 2πI
− VS sin α i + S =0 (10)
aS 2D 2) Estimated Hk+1, Qk+1 by solving Eq. (13)
together with the C- characteristic line Eqs. (10)-
xi − x S (11).
= VS − a S (11) 3) Obtained Tk+1 and ηk+1 by solving Eqs.
∆t k
(14)-(15)
where i denotes the regular x-mesh point value at 4) Improved estimation of pump run-down
speed:
location x = (i∆x) and k denotes the irregular time
level corresponding to the time at tk = Σ(∆tk). Time
57
T.S. Lee et al. / JAFM , Vol. 1, No. 1, pp. 55-61, 2008.
58
T.S. Lee et al. / JAFM , Vol. 1, No. 1, pp. 55-61, 2008.
vapour pressure are modeled. Numerical T. S. Lee (1991), ‘Numerical computation of fluids
experiments show that where gas content and air pressure transient in pumping installations with air
entrainment exist, the first peak pressure is entrainment’, International journal for numerical
amplified; the damping effect is fast; and pressure methods in fluids, Vol. 12, pp. 747-763.
surges are asymmetric with respect to static head.
The pressure transient shows a longer period for L. Jonsson (1985), ‘Maximum transient pressures
down-surge and a shorter period for up-surge. The in a conduit with check valve and air entrainment’,
paper also introduced some explanations for above Proc. Int. Conf. on the Hydraulics of pumping
effects of air entrainment on pressure transient stations, Manchester, BHRA, Cranfield, pp. 55-76.
behaviours of pumping systems. However, the
results still remained some doubts such as the J. A. Fox (1984), Hydraulic analysis of unsteady
precise physical cause of large surge damping, the flow in pipe network, Macmillan, London
physical process of gaseous diffusion in a closed
conduit subjected to unsteady flow, etc. This P. A. Dawson and J. A. Fox (1983), ‘Surge
limitation is partly due to the random nature of pressures at Riding Mill pumping station: actual
bubble nucleation, coalescence and growth in flow values and theoretical predictions’, Proc. 4th Int.
fields. Therefore, much more research in this field Conf. on Pressure Surges, Bath, BHRA, Cranfield,
is needed. pp. 427-445.
T.S Lee, H. T. Low and W. D. Huang (2004), R. J. Brown (1968), ‘Water-column separation at
‘Numerical study of fluid transient in pipes with air two pumping plants’, J. Basic Eng., ASME, 521-
entrainment’, International Journal of 531.
Computational Fluid Dynamics, Vol. 18 (5), pp.
381-391. I. S. Pearsall (1965/1966), ‘The velocity of
waterhammer waves’, Proc. Inst. Mech. Eng., 180,
Fujii, T. and Akagawa, K. (2000), ‘A study of pt. 3E, 12-20.
water hammer phenomena in a one-component
two-phase bubbly flow’, JSME Int Journal Series K. J. Whiteman and I.S. Pearsall (1962), ‘Reflux
B-Fluids and Thermal Engineering, Vol 43, Pt 3, valve and surge tests at a station’, Fluid Handling,
pp. 386-392. XIII, 248-250, 282-286.
Huygens, M., Verhoeven, R. and Van Pocke, L. K. J. Whiteman and I. S. Pearsall (1959), ‘Reflux
(1998), ‘Air entrainment in water hammer valve and surge tests at Kingston pumping station’,
phenomena’, Advances in Fluid Mechanics II, Vol Brit. Hydromech. Res. Assoc./National Engineering
21, pp. 273-282 Laboratory Joint report 1.
59
T.S. Lee et al. / JAFM , Vol. 1, No. 1, pp. 55-61, 2008.
Elevation (m)
105
100
40
Constant wave speed
Pressure head (meters)
35
Variable wave speed
30
25
20
15
10
5
0
0 50 100 150 200 250
Time (seconds)
Fig. 2- Pressure head downstream of pump
40
eps = 0.0001
Pressure head (meters)
35 eps = 0.001
30 eps = 0.01
25
20
15
10
5
0
0 50 100 150 200 250
Time (seconds)
Fig. 3- Pressure head downstream of pump with different initial air void fraction
39
Pressure head of first peak (m)
37
35
33
31
29
27
25
0 0.005 0.01 0.015 0.02
Initial air void fraction
Fig. 4- Pressure head of first pressure peak with initial air void fraction
60
T.S. Lee et al. / JAFM , Vol. 1, No. 1, pp. 55-61, 2008.
1400
eps = 0.0001
1200
Fig. 5- Wave speed downstream of pump with different initial air void fraction
1425
825
625
425
225
25
0 0.005 0.01 0.015 0.02
Initial air void fraction
Fig. 6- Wave speed downstream of pump varies with initial air void fraction
61