RESEARCH NOTE

ANALYSIS AND DESIGN OF A SIMPLE SURGE TANK

M. Azhdary Moghaddam

Civil Engineering Department, University of Sistan and Baluchestan

Zahedan, Iran, mazhdary@hamoon.usb.ac.ir

(Received: December 9, 2003 Accepted in Revised Form: October 14, 2004)

Abstract In a hydroelectric power plant or in a pumping station in order to avoid sudden large
increase of pressure due to instantaneous valve closure sometimes a surge tank is installed. The height
of surge tank is designed by the highest possible water level during the operation. The theoretical
treatment of oscillation in a surge tank is difficult because of the non-linearity of friction term in the
governing differential equation of the system. The present study attempts to find a general solution for
the surge oscillation in a simple surge tank in terms of non-dimensional parameters. Equations for the
highest and the lowest water level in the tank, which are very important in the design of a surge tank
have been found.

Key Words Surge Tank, Pressure, Surge Oscillation, Non-Linear Differential Equation


   
 
    !" # $%& '(%)   


36 7 8  .
/ 1  $ 234 4'
'5/ 4 . ) *+
 !, -&') .
/
9> ? @ , ? 4 ABC/ 1  . ) 9.) :4 ; <)  ,.3; = '
1%= .' D  
 )D # ?E4  ?4 -&') @
/ @  F') 4'
 ,
I % ) =)  )D >H  , @
/  $ 234 1/ 1D
1/G  /G)
1. INTRODUCTION
When the valve in a hydroelectric power plant is
suddenly completely closed, because of its small
inertia the water in the penstock stops almost at
once [1,3]. The water in the pipeline, with large
inertia retards slowly. The difference in flows
between pipeline and penstock causes a rise in the
water level in the surge tank. The water level rises
above the static level of the reservoir water,
producing a counter-pressure so that water in the
pipeline flows towards the reservoir and the level
of water in the surge tank drops [4].
In the absence of damping, oscillation would
continue indefinitely with the same amplitude. The
extent of damping is governed by roughness condition,
restricted orifice, and so on. The flow into the surge
tank and water level in the tank at any time during the
oscillation depends on the dimension of the pipeline
and tank and on the type of valve movement. The
IJE Transactions A: Basics
.4 ?
main functions of a surge tank are [5]:
1. It reduces the amplitude of pressure fluctuations
by reflecting the incoming pressure waves;
2. It improves the regulation characteristic of a
hydraulic turbine.
Depending upon its configuration, a surge tank may be
classified as simple, orifice, differential, or closed. A
simple surge tank is defined as a tank or shaft of
constant horizontal cross sectional area that connects
the conduit of a hydroelectric power plant for
preventing the pressure surges entering into it (Figure
1). The maximum amplitude of water level (maximum
surge) can be observed when a full load is suddenly
rejected [6].
2. MAIN CONSIDERATIONS IN THE DESIGN
OF A SURGE TANK
In order to accomplish its mission most effectively,
Vol. 17, No. 4, November 2004 - 339
 Figure 1. Simple surge tank system.
the surge tank dimensions and location are based
on the following considerations [6], [7]:
1. The surge tank should be located as close to the
power or pumping plant as possible;
2. The surge tank should be of sufficient height to
prevent overflow for all conditions of operation;
3. The bottom of surge tank should be low enough
that during its operation the tank is drained out and
admit air into the turbine penstock or pumping
discharge line; and
4. The surge tank must have sufficient cross
sectional area to ensure stability.
The height of a surge tank is governed by the
highest possible water level that can be anticipated
between the surge tank and the pipeline, also the
reservoir is considered so large that its level
remains constant [8].
3.1. Derivation of Governing Equation To
simplify the derivation of dynamic and continuity
equations that describe the oscillations of the water
level in the tank, it has been assumed that
(i) the conduit walls are rigid;
(ii) the water is incompressible; and
(iii) the effect of entrance loss in comparison
with the friction loss has been neglected.
The equation of motion is written as [9], [12] :
wV wV wy
g S0  Sf 
during its operation. All available methods are
V g (1.a)
based on a linearized resistance relationship, since wt wx wx
the resistance law flow varies as Reynolds number
and relation roughness [2]. These equations describe wV wV wy wZ wh f
approximate values of peaks and downsurges. V g g  (1.b)
wt wx wx wx wx
Integration (1.b) with respect to x between the limits
3. ANALYSIS OF SURGES IN SIMPLE x=0, x=L (see Figure 1) and simplifying, one gets
SURGE TANK
dV
L  gh f  gy 0 (2)
In a simple surge tank, there is very little head loss dt

340 - Vol. 17, No. 1, January 2005 IJE Transactions A: Basics

From continuity condition between tank and pipe, Combining 3 and 8 one gets
it can be shown that
2
fLV 2 fLD T2 dy
DT
2
dy hf (9.a)
V (3) 2gD 2P 2gD 4P dt
DP dt
With the change of flow direction, the direction of
That with use of 3 in 2 the following equation can 2
dy
be found: friction also changes. Hence occurring in
dt
2
D d2y dy dy
L T  gh f  gy 0 (4) 9.a should be split to . Thus 9.a changes to
2 dt dt
DP dt

With initial condition at t=0 fLD T2 dy dy

hf (10)
2gD 4P dt dt
y h f 0 (5.a)

2
In which 8[11]
dy DP
V0 (5.b) 0.12
dt DT

8
Equation 4 is the governing differential equation
f 64  9.5
for surge oscillation in a simple surge tank. R e 6 16

Ln 3.7H  5.74  2500
3.2. Adopted Methods for the Solution Depending
DP R e R e
0 .9

on the nature of friction loss, the following
(11)
particular cases arise:
Substituting 10 in 4 one gets
3.2.1. Frictionless Flow In this case hf = -hf0 = 0.
Thus Equation 4 changes to
d 2 y fD T2 dy dy gD 2P
  y 0 (12)
d 2 y gD 2P dt 2 2D 3P dt dt D T2 L
 y 0 (6)
dt 2 D T2 L
The initial conditions prescribed on 12 are at t=0
Solving 6 for the initial condition 5.a, 5.b one gets
f 0 LV02
y h f 0  (13.a)
DT L 2gD 2P
y V0 sint (7)
DP g
2
dy DP
Equation 7 describes sinusoidal oscillations[10]. V0 (13.b)
dt DT
3.2.2. Laminar and Turbulent Flow In this
case head loss is expressed by Darcy-Weisbach In order to reduce the number of parameters, the
equation following dimensionless groups are formed:

fLV 2 DT g
hf (8) y* y (14.a)
2gD 2P D P V0 L

IJE Transactions A: Basics Vol. 17, No. 4, November 2004 - 341

 TABLE 1. Coefficients of Equation 20.

S. No S C k1 k2 k3 Max.Error %
1 yp1* 1 (20.a) (20.b) (20.c) 4

2 tp1* S 2 (20.d) (20.e) 0.587 3

3 yt1* -1 (20.f) (20.g) (20.h) 5

4 tt1* 3S 2 1.9 1.49 0.653 4.2

5 yt01* -1 (20.1) 0.914 -0.946 4.4

6 tp2* 5S 2 3.0 1.362 0.7 4.6

DP g and the initial conditions for sudden valve

t* t (14.b) opening are
DT L
dy *
The conduit velocity may be expressed as 1 (19.b)
dt *
dy * Being non-linear it will be very difficult to find a
V V0 (15)
dt * closed form solution of 18. A numerical solution,
using a fourth order Runga Kutta method, can be
Substituting 15 in 11 one gets attempted. From this numerical solution one can
see that the various properties of the solution like
dy * H
Re R0 (16) yp1*, tp1*, yt1*, and tt1* are functions of R0, , and
dt * DP
hf0*.
H
V0 D P With varying 0 d d 0.02 the following
R0 (17) DP
Q
empirical formula for finding the mentioned values
in which R0 is the initial Reynolds number has been suggested:
defined as using non-dimensional variables 12
k3
reduces to: h k2
S C 1  f 0* (20)
2 k 1
d y* f dy dy *
2
 h f 0* *  y* 0 (18)
dt * f0 dt * dt * In which the values of S, C, k1, k2, k3 can be found
from the Table 1.
The initial conditions for sudden valve closure at
t=0 H
0.144


1.85 1  0.35
D
P
dy *
dt *
1 (19.a) k1

1.32  R 00.024  (20.a)

342 - Vol. 17, No. 1, January 2005 IJE Transactions A: Basics

 0.6
0.4
0.2
0
-0.2 0 1 2 3 4 5 6
y*

-0.4
-0.6
Eq. 20.
-0.8 Eq. 21.a
-1
Time

Figure 2. Surge Oscillation in Simple Surge Tank.

k2 1.896R 00.011 (20.b) H

0.287

k1 1.66  1.227 (20.h)

DP
k3 0.946R 00.01 (20.c)

15600 
0.57 1  1.6R 00.041 
k1 0.786  (20.d) k3 0.037
(20.i)
H 0.179 H
1 
D P
10 5  R
0   1  0.054
DP



Figure 2 shows a typical plot of surge oscillation
740
k2 1.73  for sudden valve closure obtained by solution
0.65

1  19.61 H Equation 18 using fourth order Runga Kutta
D


3162  R 0.7
0  method. Analysis of a large number of surge
P oscillation curves suggest the following empirical
(20.e) equation for h f 0*  1.5 :

H
0.016
hf 0*
1.361  0.05 y* exp Dt*   exp Et*  sinZ*t*  I
DP 2 sin I
k1 (20.f) (21.a)
1.5  R 00.035
in which
k2

0.462 1.92  R 00.024  (20.g)
H
0.083
2S
1  0.6 Z* (21.b)
t p 2*  t p1*
DP

IJE Transactions A: Basics Vol. 17, No. 4, November 2004 - 343

 S 5t p1*  t p 2* 5. NOTATIONS
I (21.c)
2 t p 2*  t p1*
DP Diameter of Conduit
DP0 Minmum Conduit Diameter
Putting t * t p1* in Equation 21.a and equating it DT Diameter of Simple Surge Tank
to y p1* one gets: DTS Stable Diameter of Tank
f Friction Factor
f0 Friction Factor for Initial Velocity
2 y p1* sin I
exp Dt p1*   exp Et p1*  (22) g Gravitational Acceleration
h f 0* h1 Free Board in Tank
h2 Cushion Level in Tank
Also by putting t * t t1* in Equation 21.a and hf Head Loss
equating it to y t1* one gets: hf0 Initial Head Loss
h f0* Non-Dimensional Initial Head Loss
2 y t1* sin I H0 Desired Head
exp Dt t1*   exp E t t1*  (23) HT Height of Tank
h f 0*
KP Cost Parameter for Conduit
KT Cost of Tank per Unit Area
The quantities of D and E can be obtained by
L Length of Conduit
solving Equations 22 and 23, simultaneously [1]. m Cost Parameter for Conduit
Figure 2 shows the result obtained by solving
P0 Desired Power
Equation 21.a and compares it with the result
obtained by numerical solution of differential Q0 Initial Discharge
Equation 18 by Runga Kutta method. The y * R Reynolds Number
R0 Initial Reynolds Number
versus t * curve can be converted to a y versus t
t time
curve via using Equations 14.a and 14.b.
t* Non-Dimensional Time Parameter
tp1 The Occurrence Time of First Peak
t p1* Non-Dimensional Time of Occurrence
4. CONCLUSION First Peak
t p2* Non-Dimensional Time of Occurrence
In the present study a general solution for surge Second Peak
oscillation in a simple surge tank and an optimal tt1 The Occurrence Time of First Downsurge
design of system has been discussed. The main t t1* Non-Dimensional Time of First Downsurge
conclusions are as follows:
V Velocity
1. Equations for maximum surge height and V0 Initial Velocity
corresponding time of occurrence have been y Height of Surge
obtained.
y* Non-Dimensional Height of Surge
2. Equations for minimum downsurge and the
corresponding time of occurrence have been yp1 Height of First Peak
developed. y p1* Non-Dimensional Height of First Peak
3. An equation for the occurrence of a second yt1 Height of First Downsurge
peak of the surge oscillation has been y t1* Non-Dimensional Height of First Downsurge
obtained. ZL Bottom Level of Tank
4. An equation for surge oscillation has been
Z0 Level of Reservoir
obtained.

344 - Vol. 17, No. 1, January 2005 IJE Transactions A: Basics

H Roughness of Conduit
K0 Steady State Efficiency
P Viscosity of Water
Q Kinematic Viscosity of Water
U Mass Density of Water
6. REFERERENCE
1. Azhdary Moghaddam, M., "Simple Surge Tank Analysis and
Design", M. E. Thesis presented at Univ. of Roorkee,
Roorkee U.P. India, in Partial Fulfilment of Required for the
Master of Engineering in Civil Engineering, (1992).
2. Chaudhary, M. H., "Applied Hydraulic Transients", Van
Nostrand Reinhold Co., New York, U.S.A., (1987).
IJE Transactions A: Basics
3. Jakobsen, B. F., "Surge Tanks", Trans. ASCE, (85), (1956),
1357-1382.
4. Jeager, C., "Engineering Fluid Mechanics", Blakie and Son
Ltd., London, U.K., (1956).
5. Marris, A. W., "Large Water Displacement in Simple Surge
Tank", J. Basic Engrg. ASME, (81), (1959), 446-454.
6. Parmakian, J., "Water Hammer Analysis", Prentice-Hall Inc.,
New York, U.S.A., (1955).
7. Pickford, J., "Analysis of Surges", Macmillan and Co. Ltd.,
bath, U.K., (1969).
8. Rich, G. R., "Hydraulic Transient", Mc Graw-Hill Book Co.
Inc., New York, U.S.A., (1951).
9. Rouse, H., "Engineering Hydraulics", John Wiley & Sons
Inc., New York, U.S.A., (1950).
10. Sutton, B. A., "Series Solution of Some Surge Tank
Problems", Proc. Instn. Civil Engrs., (16), (1960), 225-234.
11. Swamee, P. K., Jain, A. K., "Explicit Equation for Pipeflow
Problems", J. Hyd. Div. ASCE, 102(5), (1976), 657-664.
12. Wylie, E. B., Streeter, L., "Fluid Transients", Mc Graw Hill
Inc., New York, U.S.A., (1978).
Vol. 17, No. 4, November 2004 - 345