1

12
SURGE TANKS

12.1 Introduction12.2 Frictionless Analysis 12.3 Frictional Analysis 12.4 Surge Tank Operation Typical
examples. Highlights. Theoretical problems. Numerical problems.

12. 1 Introduction
Fig 12.1 shows a typical arrangement for a surge tank (or surge shaft installed in a
hydro power generation plant.

Static level y

Supply H.G.L. hf

Reservoir Supply tunnel

Area, a Surge shaft (tank)
Dia, d Area A, Dia D

Flow, Q Penstock
v Control
L valves Tail race

Ports Discharge
Turbine
Fig 12.1: Surge Tank Lower reservoir

For the arrangement shown, let:

L = length of supply tunnel
a = area of supply tunnel of diameter d
v = velocity of flow in supply tunnel
Q = Turbine demand
A = area of surge tank of diameter D
hf = frictional head loss in supply tunnel
y = rise of level in surge tank
Sudden changes in demand for turbines could cause high inertia pressures in the
supply tunnel. The function of a surge tank is thus to prevent excessive pressures
occurring when the demand is reduced, by providing a storage volume into which the
flow can pass. As the level in the surge tank rises, the back pressure increases and the
flow in the pipeline is gradually decelerated. The surge tank also provides a subsidiary
supply to ensure that sufficient water is immediately available close to the turbine when
there is a sudden increase of load.

h.m.m.
 2

12.2 Frictionless Analysis

If at any time t the surface of water in the surge tank is y above the static level and
the friction head is hf as shown in Fig 12.1,
Head opposing flow = y+ hf
where hf is +ve when the flow is into the shaft and changes to ve when the flow reverses
after level in the tank has reached its peak.
Force opposing motion of water in the supply tunnel,
= ga (y h f )
Rate of change of momentum in the supply tunnel,
dv
= aL
dt
Assuming water is incompressible and the pipe is inelastic,
dv dv
ga (y h f ) = aL (minus sign introduced since is deceleration)
dt dt
L dv
+ y hf = 0 (i)
g dt
4fL v 2
where h f = = Cv 2
d 2g
Equation (i) is referred to as the dynamic equation. At the entrance to the surge
tank, continuity of flow must apply i.e. flow along the supply tunnel at time t must equal
the flow Q to the turbines plus the flow up into the surge tank.
dy
Surface velocity into surge tank =
dt
dy
Flow into surge tank= A
dt
If v and Q are the pipe velocity and turbine demand at time t and considering
continuity of flow,
Flow down pipe = flow into surge tank + flow to turbine
dy
av = A +Q (ii)
dt
Equation (ii) is termed the continuity equation. To solve equations (i) and (ii), it is
necessary to make assumptions about the variation of friction factor and the rate of flow
reduction to the turbine. Even under total load rejection when Q=0, the variable friction
factor still prevents an analytical solution and so it is usual to use a step by step numerical
integration, preferably utilizing a digital computer.
If friction is ignored, then an approximation to the maximum surge level and its
period can be made at the point of sudden stoppage. Then, when hf and Q = 0, equations
(i) and (ii) become;
L dv
+ y = 0 , and (iii)
g dt
dy
av = A (iv)
dt

h.m.m.
 3

Differentiating equation (iv),

dv A d 2 y
= (v)
dt a dt 2
Combining equations (iii) and (v),
LA d 2 y
+y=0 (vi)
ga dt 2
d 2 y ga
or + y=0 (vii)
dt 2 LA
d2y
+ 2y = 0 (viii)
dt 2
ga
where =
LA
Equation (viii) is a linear homogeneous second order differential equation with
constant coefficients. The equation may also be recognized as that of an undamped
simple harmonic motion (SHM).
The solution of this differential equation is,
y = C1 sin t + C 2 cos t (ix)
where C1 and C2 are constants of integration.
When t = 0, y = 0, hence C2 = 0
Then y = C1 sin t (x)
T
Also when t = ,y=0
2
T
Then 0 = C1 sin
2
T
As C1 cannot be zero, sin = 0
2
T
Or =
2
2 LA
T= = 2 (xi)
ga
where T is the periodic time of surge (or oscillation).

Rise, y
T/2 T/2

Time, T

Fig 12.2 : Periodic Time of Oscillation

h.m.m.
 4

The maximum amplitude C1 is obtainable from;

av
u = C1 where u is the orbital velocity which in this case is and v is the
A
steady pipeline velocity.
u av LA
C1 = =
A ga
when t = 0, v = vo so that av = QO
Q0 A L a g
y= sin t (xii)
A a g A L

Qo A L
y max = (xiii)
A a g
The displacement curve is sinusoidal, and the surge tank water level moves
sinusoidally in relation to time as shown in Fig 12.2.

12.3 Frictional Analysis

To obtain a comparable result when friction is considered, the sudden complete
shut down must be considered as before (Q=0), but unfortunately the resulting
differential equation becomes unintegrable.
In steady state, the water level in the surge tank will be at A, an amount yi below
the reservoir static water level as shown in Fig 12.3 and
4fL v12
yi =
d 2g
= Cv in
where vi = steady state velocity, that is initial velocity

C
yf

CVn B
yi
A

Q
Fig 12.3

Negative sign is introduced since y is measured positively upwards from the

reservoir static water level.
When valve closure occurs,
dy
av = A , as in equation (iv).
dt
The dynamic equation must however be modified to allow for friction. If in time t
the water level has risen to C due to valve closure, and velocity in pipe at that instant is v,

h.m.m.
 5

the difference in level between the actual position of water surface and position it would
be in if the flow were steady at velocity v (position B) is y + Cv n . This means that the
pressure applied to the pipe by water in the surge tank is greater than that required for
steady flow by an amount y + Cv n and so the water in the pipe must experience a
deceleration given by,
L dv L dv
y + Cv n = or + Cv n + y = 0
g dt g dt
From equation (iv),
dy A dy
av = A or v =
dt a dt
n n
A dy
Cv n = C
a dt
From equation (v),
dv A d 2 y
=
dt a dt 2
n n
L A d2 y A dy
So 2
+ C + y = 0
g a dt a dt
If n = 2, then
2
d2 y A A ga dy ga
2
+ C + y=0
dt a a LA dt LA
2
d2 y gA dy ga
or 2
+ C + y=0 (xiv)
dt La dt LA
Note: By putting C=0 in the above equation, the frictionless result is obtained.
The above expression is used for a rising surge but for a falling surge, the friction
changes sign. The equation cannot be integrated but finite difference computations have
to be undertaken to estimate the magnitude of any surges.

12.4 Surge Tank Operation

The following cases arise in conjunction with surge tank operations:
(a) Partial sudden closure of a downstream valve or turbine gate

Qf
Fig 12.4

The initial steady flow is denoted by Q, and the final flow by Qf as shown in Fig
12.4. Then,

h.m.m.
 6

dy
Q QF = A
dt
dy Q Qf
or =
dt A
The dynamic equation remains unchanged so that,
L dv
y + Cv n =
g dt
(b) Sudden valve opening or turbine start up

Initial state

y Cv2

Q
Fig 12.5

Initially the water levels in the reservoir and surge tank are the same. When the
turbine starts up the flow down the penstock takes the value Q.
At a time t, then
dy
Q = A + av
dt
(Negative sign indicates that y is taking progressively larger negative values)
The flow Q is partly supplied by flow out of the surge tank and partly by flow from
the pipeline. Initially there is no flow out of the pipe and all the flow comes from the
surge tank. Finally all the flow comes from the pipeline and none from the surge tank.
The dynamic equation is,
L dv
y + Cv n = (as before).
g dt
(c) Partial opening of a valve turbine
If the initial state of the system is total shut-down there is no difference between a
sudden complete and sudden partial opening. If the initial state is that the system is
running under part load then a further partial opening up needs separate consideration.
Let the flow in the pipeline at time t be Q and the final flow Qf. Then
dy
A = Qf Q
dt
dy Q Q f
So = , and the dynamic equation is the same as before, and the result is
dt A
the same as for a partial shut-down.

(d) Slow opening or closing

h.m.m.
 7

At any time t after a valve has begun to close the flow through it is a function of
time,
Q t = Q( t )
If this valve is then substituted for Qf, the equation becomes,
dy [Q Q(t)]
=
dt A
The dynamic equation remains unaltered.
From the foregoing, it can be seen that all cases can be described by the two
equations,
dy [Q Q(t)]
(i) = , and
dt A
L dv
(ii) y + Cv n =
g dt
For a sudden closure, Q(t) = 0
For a partial closure, Q(t) = Qf
For sudden opening, Qi = 0 and Q(t) = Qf
and so on.

Example 12.1: In a hydro-electric scheme, the supply pipeline is 1.2m in diameter and
has a resistance coefficient f=0.01. At 150 m along the pipeline from the reservoir, there
is an open surge tank, 3.6m in diameter with no restriction. The steady flow to the turbine
is 2.27m3/s.
(a) Develop the basic equations for mass oscillation in the surge tank due to a sudden
change of flow.
(b) Show by giving a few stages of step-by step integration how to estimate the
maximum rise of water level in the surge tank for a sudden full flow rejection.

Solution
(a) From equation (i),
L dv 4fL v 2
+ y h f = 0 , where h f = = Cv 2
g dt d 2g
4 0.01 150v 2
= = 0.255v 2
1.2 2 9.81
From equation (ii),
dy
av = A + Q,
dt

where a = 1.2 2 = 1.13m 2
4

A = 3.6 2 = 10.17 m 2
4
Substitute values in equation (i):
150 dv
+ y 0.255v 2 = 0 (1)
9.81 dt

Substitute values in equation (ii):

h.m.m.
 8

dy
1.13v = 10.17 +Q
dt
dy
v=9 + 0.866Q (2)
dt
Equations (1) and (2) are the basic equations for this problem.
(b) An approximate estimate of the maximum surge level and the time it occurs
can be obtained by neglecting friction.
Qo A L
y max =
A a g
3 2 2
Q o =2.27m /s, A= 10.17m , a= 1.13m , L=150m
2.27 10.17 150
ymax = = 2.62m
10.17 1.13 9.81
Periodic time of surge,
AL 10.1`7 150
T = 2 = 2 = 73.8s
ag 1.13 9.81
The surge follows a sign wave and the maximum rise in level occurs at
1
T = 18.45s after start.
4
More Accurate Analysis
This can be made using step-by-step integration of equations (1) and (2).
For total load rejection, Q = 0
Replace dy, dt and dv by finite integrals y, t and v so that equation (2)
becomes,
y
v=9 (3)
t
and equation (1) for upsurge is,
v
y = 0.255v 2 + 15.3 (4)
t
At start, when t = 0
Q O 2.27
vo = = = 2.01m / s
a 1.13
y o = 0.255v 2 = 0.255 2.012 = 1.03m
Working in steps of t =5s, work out mean values of v and y. If vi and yi are the
values of v and y at the beginning of each step, mean values for the step are:
1
vm = vi + v (5)
2
1
y m = y i + y (6)
2
The procedure for each step is as follows:
(i) Estimate y , then calculate v m from equation (3):
y
vm = 9
t

(ii) Using this value of v m , calculate

h.m.m.
 9

v = 2(v m vi ) , based on Eq. (5).

Then from equation (6), evaluate
v
y m = 0.255v 2 + 15.3
t
(iii) Compare the values of y m from step (ii) with the value of y m from equation
(6). If the two values do not agree, re-estimate y and repeat the step.
If agreement is satisfactory, proceed to the next step with initial conditions
y i + y and v i + v from previous step. The working is carried out as shown
in Table 12.1. As the peak value of y is approached, the time interval t is
reduced.

Table 12.1

ym=-{0.255vm2+15.3(v/t)}
vm= 9 y/t (m/s)
Estimated y

15.3(v/t)
v = 2(vm -vi)

ym=yi+y
{ Eq. (4)}
t seconds

0.255vm2

{Eq.(6)}
vi (m/s)
yi (m/s)

Remarks
t

1 2 3 4 5 6 7 8 9 10 11 12
y v
Using a 5 seconds interval, vm= 9 = 1.8y and 15.3 = 3.06v
5 5
0 5
-1.030 2.01
1.10 1.980 -0.060 -0.184 1.000 -0.816 -0.480 ym values unequal
1.05 1.890 -0.240 -0.734 0.911 -0.177 -0.505 ym values unequal
1.075 1.935 -0.150 -0.459 0.955 -0.496 -0.493 ym values equal
5 5 0.045 1.86
0.92 1.656 -0.408 -1.248 0.699 0.549 0.505 ym values unequal
0.924 1.663 -0.394 -1.204 0.705 0.499 0.507 ym values unequal
0.923 1.661 -0.397 -1.215 0.704 0.511 0.507 ym values equal
10 5 0.968 1.463
0.650 1.17 -0.586 -1.793 0.349 1.444 1.293 ym values unequal
0.660 1.188 -0.550 -1.683 0.360 1.323 1.298 ym values unequal
0.662 1.192 -0.543 -1.661 0.362 1.299 1.299 ym values equal
15 5 1.630 0.920
0.330 0.594 -0.652 -1.995 0.090 1.905 1.795 ym values unequal
0.340 0.612 -0.616 -1.885 0.096 1.789 1.800 ym values unequal
0.339 0.610 -0.620 -1.896 -0.095 1.801 1.800 ym values equal
20 2 1.969 0.30
y v
2 Peak level is close; reduce t to 2 seconds: vm= 9 = 4.5y ; 15.3 = 7.65v
2 2
0.040 0.180 -0.240 -1.836 0.008 1.828 1.989 ym values unequal
0.037 0.167 -0.267 -2.043 0.007 2.036 1.988 ym values unequal
0.038 0.171 -0.258 -1.974 0.007 1.967 1.988 ym values equal
22 2.007 0.042 This is effectively the maximum rise of water level (yi seems to stabilize) = 2.007m at
22 s from closure of valve.

Notes: (i) Column 3: yi (first value) = yo=-0.255v2=-0.255(2.01)2 =-1.030m

h.m.m.
 10

Qo
(ii) Column 4: vi (first value) = =2.01m/s
a
(iii) Subsequent value of yi =-1.030+ 1.075 = 0.045m
(iv) Subsequent value of vi =2.01-0.150=1.86m/s

Example 12.2: A surge chamber 10m in diameter is situated at the downstream end of a
low pressure tunnel 10km long and 3m in diameter. At a steady discharge of 36m3/s, the
flow to the turbine is suddenly stopped by closure of the turbine inlet valves. Determine
using the approximate method the maximum rise in level in the surge chamber, and its
time of occurrence.

Solution
Qo A L
y max =
A a g
3
Q o =36m /s, L=10 000m, A = 10 2 = 78.54m 2 , a = 32 = 7.069m 2
4 4
36 78.54 10 000
ymax = = 48.78m
78.54 7.069 9.81
Periodic time of surge,
LA 10 000 78.54
T = 2 = 2 = 668.67s
ga 9.81 7.069
The surge follows a sign wave and the maximum rise in level occurs at
1
T = 167.17s from start.
4

h.m.m.