chapter 12

Hydraulic Turbine Prime Movers

12.1 Inhuduction
The generation of hydroelectric power is accomplished by means of hydraulic turbines that
are directly connected to synchronous generators. Four types of turbines or water wheels are in
common use. The three most common are the impulse or Pelton turbine, the reaction or Francis
turbine, and the propeller or Kaplan turbine. A fourth and more recent development is the Deri-
az turbine, which combines some of the best features of the Kaplan and Francis designs. All of
these types make use of the energy stored in water that is elevated above the turbine. Water to
power the turbines is directed to the turbine blading through a large pipe orpenstock and is then
discharged into the stream or tailrace below the turbine. The type of turbine used at a given lo-
cation is based on the site characteristics and on the head or elevation of the stored water above
the turbine elevation.

12.2 The Impulse Turbine

The impulse or Pelton wheel is generally used in plants with heads higher than 850 feet
(260 meters), although some installations have lower heads. One plant, at Bucks Creek in Cali-
fornia, has a static head of 2575 feet (785 m) and another in Switzerland has a head of over
5800 feet (about 1800 m).
Impulse turbines are often installed on a horizontal shaft with the generator mounted beside
the turbine. Some designs have two turbines on a shaft with a generator between them and are
called “double-overhung” units. The turbine wheel is spun by directing water from nozzles
against the wheel paddles and using the high momentum of the water to drive the wheel. Figure
12.1 shows a double-overhung unit with a single nozzle for each wheel. Occasionally, several
nozzles are directed toward each wheel. A stripper, also shown in Figure 12.1, is used to clear
water from the bucket as it moves upward, thereby increasing the efficiency of the unit.
Speed regulation of the impulse turbine is accomplished by adjusting the flow of water
through the nozzle by means of a needle that can be moved back and forth to change the size of
the nozzle opening. This arrangement is shown in Figure 12.2 (a) and is seen to be similar to the
familiar garden hose nozzle. This needle adjustment is used to make small, steady changes in
water flow and power input. However, since the impulse wheel is used in plants having high
heads and long penstocks, it is not advisable to use the nozzle to cut off the water jet abruptly.
The reason for this is that a sharp cut-off in flow causes a pressure wave to travel back along the
penstock causing possible damage due to water hammer. Thus, another means must be found to
divert the water stream away from the wheel while the nozzle is closed slowly. One way this is
accomplished is by mechanically deflecting the water stream by means of a jet deflector as
 Hydraulic Turbine Prime Movers 485

Fig. 12.1 A double-overhung impulse wheel.

shown in Figure 12.2(b). Thus, the governor of an impulse wheel will control the nozzle for
normal changes, but must recognize a load rejection by quickly moving the jet deflector.
In an impulse turbine, the total drop in pressure of the water occurs at the stationary nozzle
and there is no change in pressure as the water strikes the bucket. All of the energy input to the
shaft is in the form of kinetic energy of the water, and this energy is transformed into the me-
chanical work of driving the shaft or is dissipated in fluid friction. Ideally then, the water veloc-

Fig. 12.2 Impulse wheel nozzle and deflector arrangements.

A86 Chapter 12

ity is reduced to zero after it strikes the turbine buckets. Actually, a small kinetic energy re-
mains and is lost as the deflected water is directed downward to the exit passageway.
The power available at the nozzle is given by the formula

P , = - wHQ hp (12.1)
550
where
P, = power availble at the nozzle, hp
W = weight of one cubic foot of water = 62.4 lbm/ft3
Q = quantity of water, ft3/s
H = static or total head, ft
Recall that 550 l b d s is equal to one horsepower.
If 77, is the turbine efficiency, the shaft power may be written as
HQT,
P,,= -hp (12.2)
8.8
where the maximum efficiency is usually 80 to 90% [ 13. The quantity of water depends on the
water velocity, the head, and a nozzle coefficient. It is also restricted by the mean river or
stream flow, which is dictated by nature. For a given design, we can compute
Q =AVft3/s (12.3)
where
A =jet area, ft2
V = jet velocity, ft/s
Then
v=cv?@ft/s (12.4)
where
g = 32.2 ft/s2
h = net head at nozzle entrance, ft
C = nozzle coefficient, usually 0.98
If we assume that
h=kH
for a given situation,where k is a constant, then we may write
Ps= k,H3I2 (12.5)

12.3 The Reaction Turbine

In the impulse turbine, the high pressure in the penstock at the nozzle is changed to mo-
mentum so that no pressure drop is experienced at the turbine. In the reaction turbine, however,
there is only a partial pressure drop at the nozzle, the remainder taking place in the rotating run-
ner. Thus, water completely fills the cavity occupied by the runner, flows across this pressure
drop, and transfers both pressure energy and kinetic energy to the runner blades. Since so much
of the turbine blading is active in this energy transfer, the diameter of the reaction turbine is
smaller than an impulse turbine of similar rating.
Most reaction turbines in use today are of a radial inward-flow type known as the “Francis”
turbine after James B. Francis, who designed the first such water wheel in 1846. In these turbine
designs, water under pressure enters a spiral case surrounding the moving blades and flows
through fixed vanes in a radial inward direction. The water then falls through the runner, exert-
 Hydraulic Turbine Prime Movers 407

ing pressure against these movable vanes and causing the runner to turn. The generator is usual-
ly directly connected to the runner shaft as shown in Figure 12.3.
Reaction turbines are classed as radial flow, axial flow, or mixed flow according to the di-
rection of water flow. In radial flow, the water flows perpendicular to the shaft. In axial flow the
stationary vanes direct the water to flow parallel to the shaft. Mixed flow is a combination of ra-
dial and axial flow.
Reaction turbines are installed either in a horizontal or vertical shaft arrangement, with the
vertical turbines being the most common. It is a versatile design, being applicable to installa-
tions with heads as high as 800 feet (244 m) and as low as about 20 feet ( G 6 m).
The control for a reaction turbine is in the form of movable guide vanes called wicket gates
through which the water flows before reaching the runner. Positioning these vanes can cause the
water to have a tangential velocity component as it enters the runner. For one such position,
usually at 80 to 90% of wide open, the runner will operate at maximum efficiency. At any other
wicket gate setting, a portion of the energy is lost due to less efficient angling of the water
streamline. Although the wicket gates are close-fitting, they usually leak when fully closed and
subject to full penstock pressure. Thus, a large butterfly valve is often installed just ahead of the
turbine case for use as a shut-down valve.
The draft tube is an integral and important part of the reaction turbine design. It serves two
purposes. It allows the turbine runner to be set above the tailwater level and it reduces the dis-
charge velocity, thereby reducing the kinetic energy losses at discharge. The large tube with the
90" bend just below the runner in Figure 12.3 is the draft tube.
The importance of the draft tube is evident when the energy of water leaving the runner is
considered. In some designs, this energy may be as high as 50% of the total available energy.
Without the draft tube, this kinetic energy would be lost. With the draft tube constructed air-tight,
however, a partial vacuum is formed due to the fast-moving water. This low pressure tends to in-
crease the pressure drop across the turbine blading and increase the overall efficiency.
One of the important empirical formulas used in waterwheel design is the specific speed
formula.

(12.6)

Fig. 12.3 A typical vertical shaft reaction turbine arrangement.

488 Chapter 12

Table 12.1 Typical Specific Speeds for Watenvheels

Type of Wheel NS max Ns
Impulse 0 to 4.5 10
Reaction 10 to 100 150
Propeller 80 to 200 250
Deriaz 10 to 100

where
N = speed in rpm
H = head in feet
Ps= shaft power in hp
This quantity is the speed at which a model turbine would operate with a runner designed
for one horsepower and at a head of one foot. It serves to classify turbines as to the type appli-
cable for a certain location. As a general guide, then, we say that the specific speeds given in
Table 12.1 are applicable.
Under this classification, an impulse turbine is a low-speed, low-capacity (in water vol-
ume) turbine and the reaction turbine is a high-speed, high-capacity turbine. The same formulas
(12.1) to (1 2.5) used in conjunction with the impulse turbine also apply for the reaction turbine.
For (12.4), the value of C is about 0.6 to 0.8 and this value usually decreases for turbines with
higher values of Ns.
The control of a reaction turbine is through the movable wicket gates. These are deflected
simultaneously by rotating a large “shifting ring” to which each gate is attached. The force re-
quired to move this assembly is very large and two servomotors are often used to rotate the ring,
as shown in Figure 12.4.

Fig. 12.4 Wicket gate operating levers and position servomotors. Figure courtesy F. R. Schleif, Electric Power
Branch, Bureau of Reclamation, U.S. Department of the Interior. USBR photo by C. W. Avey.
 Hydraulic Turbine Prime Movers 489

The machine shown in Figure 12.4 is one of the generators at the Grand Coulee Dam Pow-
erhouse in Washington State. It shows the wheel pit of a 165,000 horsepower turbine generator.
The two rods are connected to power servomotors and operate to rotate the shifting ring, there-
by changing the wicket gate position of all gates.
A second control device used in reaction turbines is a large bypass valve, which is actuated
by the shgting ring.If load is rejected and the wicket gates are driven closed very quickly by the
governor servomotor, the pressure regulator is caused to open and does so very rapidly. This
prevents the large momentum of penstock water from hammering against the closed wicket
gates. The pressure regulator then closes slowly to bring the water gradually to rest.

12.4 Propeller-Type Turbines

The propeller-type turbine is really a reaction turbine since it uses a combination of water
pressure and velocity to drive the shaft. It employs water velocity to a greater extent than the
Francis turbine. It also has a higher specific speed, as indicated in Table 12.1.
Three types of propeller turbines can be discussed. The fixed blade or Nagler type was de-
veloped in 1916 by F. A. Nagler. It operates at a high velocity and operates efficiently only for
fixed head and constant flow applications.
A few years later, in 1919, Kaplan developed the adjustable blade propeller turbine shown
in Figure 12.5. This design has the advantage of fairly high efficiency over a wide range of head
and wicket gate settings. Adjustments of wicket gate setting and blade angle can both be made
with the unit running. This permits optimization of turbine efficiency over a wide range of head
and load conditions.
Kaplan turbines are used at locations with heads of 20 to 200 feet (about 15 to 150 m).
Compared to the Francis turbines, the Kaplan units operate at higher speeds for a given head
and the water velocity through the turbine is greater, leaving the runner with a fast swirling mo-
tion. Thus, the draft tube design is important in Kaplan turbine applications.

12.5 The Deriaz Turbine

The Deriaz turbine is a more recent development in reaction turbine design and incorpo-
rates the best features of the Kaplan and the mixed-flow Francis designs. It is essentially a pro-
peller turbine with adjustable blades. The blades are contoured similar to the Francis blading
and are set at 45 degrees to the shaft axis rather than 90 degrees as in the Kaplan turbines. These
differences are illustrated in Figure 12.6, where the blades are identified by the letters A and the
direction of water flow by the letter W.
Wicket gates are generally not used with a Deriaz turbine and control is maintained by
blade adjustment only. The Deriaz turbine has the capability of operating at high turbine effi-
ciency over a wide range of loadings, as shown in Figure 12.7. Thus, this design is well suited
for situations requiring large variations in loading schedules.

12.6 Conduits, Surge Tanks, and Penstocks

It is assumed that any hydroelectric generation site has a supply of elevated water from
which water may be drawn to power the turbine. The selection of sites and construction of
dams, spillways, and the like are important, but are beyond the scope of this text. Many excel-
lent references are available that discuss these important items [5, 61. We will assume that a
reservoir of water exists and is large enough in capacity that, during periods of interest for con-
trol analysis, the head is constant. That is to say, the water source is an infinite bus.
From the reservoir, water is drawn from an area called the forebay into a couduit or large
pipe, and flows to the turbine as shown in Figure 12.8. In some cases, a relatively level section
490 Chapter 12

Fig. 12.5 The Kaplan propeller turbine.

of pipe, called the conduit, is necessary to move the water to a point where it begins a steep de-
scent through the penstock to the turbine. As the water flows through this conduit and penstock
at a steady rate, a head loss develops, similar to the voltage drop in a nonlinear resistor. The hy-
draulic gradient in Figure 12.8 represents the approximate profile of the head, measured in feet,
as a hnction of distance from forebay to turbine. Under steady-flow conditions, this head loss at
the turbine is

hL = H - h = kQ" (12.7)

where
hL= head loss, feet
H = static head, feet
h = effective head at the turbine, feet
k = a constant corresponding to pipe resistance
 Hydraulic Turbine Prime Movers 491

(a) The Francis Runner

(b) The Kaplan Runner

(c) The De& Runner

Fig. 12.6 Comparison of reaction turbine runners.

Q = flow rate, ft3/s

n = a constant, where 1 5 n 5 2
Thus, when the flow is steady, the head loss will be directly proportional to the length of
pipe, as indicated in the figure.
One of the serious problems associated with penstock design and operation is that of water
hammer. Water hammer is defined as the change in pressure, above or below normal pressure,
caused by sudden changes in the rate of water flow [ 6 ] .Thus, following a sudden change in
load, the governor will react by opening or closing the wicket gates. This causes a pressure
wave to travel along the penstock, possibly subjecting the pipe walls to great stresses. Creager
[ 6 ] gives a graphic example of this phenomena as shown in Figure 12.9. Suppose the load on
the turbine is dropped suddenly. The turbine-governor reacts to this change by quickly moving
the wicket gates toward the closed position and, because of the momentum built up by the pen-
stock water, the hydraulic gradient to changes from the normal full load gradient A-C, to the
positive water-hammer gradient, A-D. This supernormal pressure is not stable, and once the
wicket gate movement stops, gradient A-D swings to A-E and oscillates back and forth until
damped by fhction to a new steady-stateposition.
492 Chapter 12

100 4
Denaz

Impulse
Kaplan

Francis
N, = 50
Francis
N , = 100
Fixed
Propeller

I 1 I I I >
0 ‘
0 20 40 60 80 100
% of Full Load

Fig. 12.7 Turbine efficiency as a hnction of load.

A sudden increase in load, accompanied by wicket gate opening has just the opposite ef-
fect. Thus, not only must the penstock be well reinforced near the turbine, but it must be able to
withstand these shock waves all along its length.
Examining this phenomenon more closely, reveals that it is much like the distributed para-
meter transmission line. The (closing) wicket gate can be thought of as a series of small step
changes in gate position. Each step change causes a positive pressure wave to travel up the pen-
stock to the forebay and, upon reaching this “open circuit,” it is reflected back as a negative

----_ Static Hydraulic Gradient

-- -
Tailrace
Fig. 12.8 A typical conduit and penstock arrangement.
 Hydraulic Turbine Prime Movers 493

wl
\\ I
Penstock

-
I i k..-
Wicket Gates f _i

ne
Tailrace

Fig. 12.9 Hydraulic gradient following a loss of load.

pressure wave of almost the same magnitude. The time of one “round trip” of this wave is called
the critical time, p, which is defined as
2L
p = - seconds (12.8)
a
where
L = length of penstock, feet
a = pressure wave velocity, ft/s
For steep pipes, the wave velocity is approximately
4675 ft/s
a= (12.9)
1 + (d100e)
where
d = pipe diameter, inches
e = pipe wall thickness, inches
Pressure wave velocities of 2000 to 4000 feet per second are not uncommon.
The change in head due to water hammer produced by a step change in velocity has been
shown to be [6]

(12.10)

where
hA= change in head, feet
vA = change in velocity, W S
g = acceleration of gravity, ft/s2
and a is the pressure wave velocity as previously defined. Equation 12.10 is the hdamental
equation for water hammer studies. Note that to keep water hammer to a low value, vA must be
494 Chapter 12

kept small either by using a pressure regulator or by introducing intentional time lag in the gov-
ernor. The introduction of time lags are particularly troublesome for interconnected operation as
this contributes to tie-line oscillation [7].
Usually, the time for closure of the wicket gates of a hydraulic turbine is much greater than
p of equation (12.8). Suppose, however, that the gate is opened by only a small amount, such
that it can be closed in a time p . In such a case, the pressure rise can be greater than that due to
closure from full gate to zero. For this reason, p is usually considered the critical governor time.
From the above, we see that water hammer, both positive and negative, can be a serious
problem in penstock design. It may require that penstocks be built with much greater strength
than would ordinarily be necessary. It may also cause violent pressure oscillations, which can
interfere with turbine operation. The pressure regulator is helpful in controlling positive water
hammer as it provides relief for the pressure buildup due to closing of the gates. However, it is
of no help in combating negative water hammer.
A device often used to relieve the problems of both positive andnegative water hammer is
the surge tank, a large tank usually located between the conduit and penstock, as shown in Fig-
ure 12.10. To be most effective, the surge tank should be as close to the turbine as possible but,
since it must also be high enough to withstand positive water hammer gradients without over-
flowing, it is often placed at the top of the steep-descent portion of the penstock, as shown in the
figure. Sometimes an “equalizing reservoir” is constructed to serve as a surge tank for large in-
stallations and may actually be cheaper and more beneficial. This is due to the general rule that
the larger the tank area, the smaller the pressure variation [6].
Surge tank dimensions are important. The tank must be high enough so that in no case is air
drawn into the penstock. Letting y denote the maximum surge up or down in feet (measured
from the reservoir level for starting, from a distance below this equal to the friction head for
stopping) we have [ 5 ]
y= (gA
aLv% + P y 2
(12.11)

where
a = conduit area, ft2
L = conduit length, ft

Surge
Tank
-.
I

--
Forebay

Tailrace

Fig. 12.10 Conduit and penstock with a surge tank.

 Hydraulic Turbine Prime Movers 495

v,, = velocity change, Ws

g = 32.2 fus2
F = friction head, ft
A = area of surge tank, ft2
Barrow [ 5 ] also gives a formula for the time interval that elapses between turbine load
change and the occurrence of the maximum surge as

(12.12)

where
c = coefficient of fiction
cv2 = q = flow in ft3/s

The factor F in (12.10) is important since it represents the friction that eventually damps
out oscillationsfollowing a sudden change. Since damping is desirable, it is sometimes advanta-
geous to add hydraulic resistance at the surge tank opening to produce a choking effect. This is
done in two ways: by placing a restricted orifice between the tank and the penstock, or by con-
structing a “differential surge tank.”The differential surge tank, shown in Figure 12.11, consists
of two concentric tanks: an inside riser tank of about the same diameter as the penstock and an
outer or surge tank of larger diameter with a restricted passage connecting it to the penstock.
Because of this restriction, the water level in the outer tank is independent of the accelerating
head and the head acting on the turbine. These heads are determined by water in the riser tank,
which acts like a simpler surge tank with small diameter. The diameter of the differential surge
tank is about one-half that of a simple surge tank. The riser diameter is usually the same as that
of the penstock.
The damping effect due to the added friction of the differential surge tank is shown in Fig-
ure 12.12, where the surge is compared for two types of tank design [6].Note the relatively long
period (about 300 seconds, or five minutes) of the surge. This surge would be due to a sudden
increase in load, where the turbine wicket gates are opened at time t = 0. Note that an accelerat-
ing head is created, which increases steadily for about 80 to 85 seconds, at which time the flow

Surge Riser
I
A - I
- -
___ -

Tailrace

Fig. 12.1 1 The differential surge tank.

496 Chapter 12

Differential:
5

Q5
v1 15

20

25 -.: I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
1 I
I
I
I
I
I
I
I
1 %
0 50 100 150 200 250 300 350
Time in seconds
Fig. 12.12 Comparison of surges in simple and differential surge tanks.

of water from that tank ceases. In the differential tank,the accelerating head is established very
fast, but not so fast as to prevent the governor from keeping up with the change.
In the discussion of a technical paper [SI, deMello suggests a lumped parameter electric
analog of the hydraulic system, including conduit, surge tank, penstock, and turbine [9]. Figure
12.13 shows this analog, where head is analogous to voltage, volumetric flow is analogous to
current, and the turbine is represented by the variable conductance, G.
With water being considered incompressible, the inertia of water in the penstock and con-
duit are represented by inductances L,and L2, respectively (series resistance could be added to
represent hydraulic resistance). If the effect of water wheel speed on flow is neglected, the tur-
bine can be simulated by G or GA,where a change in gate setting is under consideration. The
surge tank behaves much like a capacitor as it tends to store water (charge) and release it when
the head (voltage) at the turbine falls. (How could a differential surge tank be represented?)

Conduit Penstock

I I I
Fig. 12.13 Electric analog of the hydraulic system.
 Hydraulic Turbine Prime Movers 497

If linearized equations about a quiescent operating point are written we have, for the head
at the reservoir described in the s domain,

(12.1 3)

where

Also

(12.14)

From the square root relationship between flow and head

Q=GG (12.15)
we write
il = G G (12.16)
Combining, we get

2(GA/GO)v10
ilA = (12.17)
2vo
- s(L,
+ + L,)(1 + LCS2)
i0 1 + L2C2S2
Now, assume a change in turbine power at constant efficiency or
PA = vIOiIA + ilOVIA

Po$(?- s(L, + L2) + ?L2CS2 - L,L2Cs3)

10
- (12.18)
vo
-+ +L2) +voL2cs2I L1L2cs3
i0 2 10 2
When the surge tank is very large, C is large and (12.18) reduces to the so-called water-
hammer formula

Po?( 1- ks) (12.19)

1 +-s
Jh
2RO
where
VI0
Ro= 7 (12.20)
10

Then (12.19) may be written as

PA = (12.21)
TW
1 +-s
2
490 Chapter 12

where [9]
,'Z = water starting time = 1 second (12.22)
Furthermore, as pointed out by deMello [9], when the tunnel inertia is great, or L, is large,
then (12.19) becomes

(12.23)

These results are not greatly changed by considering the conduit and penstock as a distrib-
uted parameter system.

12.7 Hydraulic System Equations

The hydraulic system and water turbine transfer functions have been thoroughly analyzed
by Oldenburger and Donelson [8]. This excellent description is based on a rigorous mathe-
matical analysis and is supported by substantial experimental evidence to testify to its validi-
ty.
As shown in the previous section, the flow of water through a conduit is analogous to an
electric transmission line in which head is analogous to voltage and volumetric flow rate is anal-
ogous to current. This is easily seen when the partial differential (wave) equations for a uniform
pipe with negligible friction are examined. For the uniform pipe, we write
du =--cy- dh
-
dx at
du dh
(12.24)
dt =-gdx
-
where
u = water velocity, Ws
x = distance along pipe, ft
h = head, R
-cy =a constant = p g ( k + X)
p = density of fluid
g = acceleration of gravity
K = bulk modulus of elasticity of fluid
r = internal pipe radius
f = pipe wall thickness
E.= Young's modulus for the pipe
Equation (12.24) should be compared to the equations of the transmission line, which can
be written as follows:

--d i= C-
a v
+ GV (12.25)
ax dt
The similarity for the lossless case should be obvious.
 Hydraulic Turbine Prime Movers 499

Now, let us define the following:

H = H(s, x) = L[h(t,x)]
u = U(S,x ) = L[u(t,x ) ] (12.26)
We may write the Laplace transform of (12.24) with the result, assuming zero initial condi-
tions,

dx
dH 1
- = _ - su (12.27)
h g
The solution of (12.27) may be shown to be
u = K e-sx/a + K2e+sda
I
H = K3e-sda + K4e+.Sda (12.28)
This result can be written in hyperbolic form as
sx sx
U = C, cosh - + C2 sinh -
a a
sx sx
H = C3cosh - + C4sinh - (12.29)
a a
where
a= = wave velocity (12.30)
These results may be simplified by eliminating of the arbitrary constants subscripted by 3
and 4. With this simplification,we have [8]
,y = ~ , ~ - s x +
/ a~ ~ ~ + s x / a

(12.31)

or
sx sx
U = CI cosh - - C2sinh -
a a

H=-- c1 cash -
sx c,
--
sx
(12.32)
Gi a 6sinh~
Note we may apply (12.31) or (12.32) to any cross section of pipe such as I or I1 of Figure
12.14, or any arbitrary cross section i. Thus, in (12.31) and (12.32) we may subscript all x's
with a numeral (I, 11, or i) to indicate the particular section under study. This helps in evaluating
the constants C,, C2,K,, and K2 as they depend on boundary conditions. For example, we may
write
S S
C, = U, cosh -XI + sinh -4
a a
S S
C2= - 6 g H , cosh -X,- U, sinh -XI (12.33)
a a
500 Chapter 12

Fig. 12.14 A view of an arbitrary pipe section selected for study.

We may then write (12.32) as, for the section at 11,

S S S S
U, = U, cosh -XI cosh -&, + *HI cosh -X,,sinh -X,
a a a a
S S S S
- 6 g H , sinh -XI, cosh -XI - U, sinh -X, sinh -X, (12.34)
a a a a
Now, let
x,=o
X, = L = length of pipe (12.35)
Then, (12.33) and (12.34) become
c,= u,
C2= --HI (12.36)
and
Ulr= U, cosh Tp - agH, sinh Tp
UI
HI, = -- sinh Tp + HI cosh Tp (12.37)

where
L
T, = - = elastic time (12.38)
a
Now, since
q=AU (12.39)
where
q = volumemetric flow rate, R3/s
A = pipe cross sectional area, ft2 (12.40)
then we may write

or, simply
Q=AU (12.42)
and this applies at any section such as I or 11. Thus, we convert the U equation to a Q equation
and rewrite (12.37) as
 Hydraulic Turbine Prime Movers 501

I
QII= QI cosh Tp - - sinh Tp
ZO
HII= -ZoQI sinh Tp + HI cosh Tp (12.43)

where
1
z --= the “characteristic” impedance (12.44)
O- A 6
From the time-domain translation theorem of Laplace transform theory we write
e-bsF(s)= L[u(t - bMt - b)] (12.45)
We readily conclude that the Laplace transform of the following differential equation may
be written:
L[(sinh T,p)f(t)] = F(s) sinh Tes (12.46)
for T, > 0 andf(t) = 0 when t < Te and where we use the notationp = d/dt. Similarly, we also
write
L[(cosh T,p)f(t)] = F(s) cosh Tes (12.47)
forf(t) = 0 when t < T,.
From these relations, we conclude that the second item in (12.43) is the Laplace transform
forf(t) when t < T,.
We can see that (12.43) is the Laplace transform of the equations
1
411 = (cash TeP)qI - -(sinh Tep)hI
ZO
hII = -Zo(sinh T,p)qI + (cosh T,p)hI (12.48)

where
4x0, t ) = h,(O, t) = 0 for t > T,
Now note that (12.46) can be rearranged and hyperbolic identities used to write
1
QI= QIIcosh Tes + -HIIsinh T,s
ZO
HI = ZoQIIsinh T,s + HIfcosh Tes (12.49)

and in the time domain this equation pair becomes

(12.50)

where
qI,(L, t ) = hI,(L, t) = 0 for t < T,
Now, we rearrange (12.49) and subsequently (12.50) to write the hybrid equation pair
1
41 = (cash TeP)q/I + Z,(si& TeP)h,

= (sech Tep)hI - Z~(tanhTePkll (12.51)

502 Chapter 12

Equations (12.51) may be evaluated by expanding the hyperbolic differential operators in

an infinite series. We recall that

(12.52)

and if this series converges rapidly, we may write approximately

e-'&f(t) = (1 - Tep)f(t) (12.53)
or, if more accuracy is require, we may add more terms. In a similar way, we may expand the
hyperbolic terms by the expansions

If these sequences in u = Tep converge rapidly, we may write for the first of equations
(12.5 1)

(12.54)

We also note that equations (12.5 1) are linear in both q and h such that, if we define

(12.55)
and write new equations in terms of the A-quantities, the new equations will be identically the
same as (12.51).
The head loss due to friction has been shown to be proportional to q2.Thus, the head equa-
tion is, from (12.51) and including a friction-lossterm

2 (sech TeP)hI - z o ( d TeP)qIl- k; qk (12.56)

This nonlinearity is removed by the approximation (12.55), or

hIIA = (sech Tep)h,A- zO(tanh TeP)qIIA- k2q11A (12.57)

where

k2 = 2k; 4110 (12.58)

We may also write (12.51) and (12.57) in per-unit terms by dividing through by a base
quantity. Let
Base q = qo

Base h = ho
Then, in per-unit terms, (12.51) becomes
1
41 = (cash TeP)qII + -(si& Tep)hll
ZO
(12.59)
 Hydraulic Turbine Prime Movers 503
where we define
hI
per unit hI = -
h0
hII
per unit h, = -
ho
41
per unit q1= -
40

411
per unit qII= -
40
zo40
per unit Zo= Z, =- (12.60)
h0
We need not use any special symbol to indicate whether these are per-unit or system quan-
tities as the equations are identical (except for Zoand Z,). In what follows, we will assume:

1. All flows and heads are deviations from the steady state, but we will avoid using the A
subscript for brevity.
2. All values are per unit.

12.8 Hydraulic System Transfer Function

We now apply the equations of Section 12.7 to typical hypothetical situations and derive
transfer functions for the hydraulic system. In so doing, we are interested in dynamic oscilla-
tions about some quiescent operating point. Partial derivatives of nonlinear relationships are as-
sumed to be derived at the quiescent point or Q-point.
The results of this section and the assumptions made have been verified for at least one
physical case as recorded in [8]. Verification was checked by the frequency-response method
[8, 101, wherein the wicket gates are oscillated at a range of frequencies and measurements tak-
en to determine the system Bode diagram. We will not dwell on this technique except to ac-
knowledge that experimental verification has been checked by others.
It has been observed in physical situations that when the wicket gates are oscillated at low
frequencies, the levels in the riser tank and surge tank are practically the same. Also, when the
frequency of oscillation is high, the levels in both tanks are practically constant as the water in-
ertia prevents it from responding to rapid changes. Thus, we assume that the levels in riser and
surge tanks are identical, or
h, = h, (12.61)
where
h, = surge tank head, per unit
h, = riser tank head, per unit
Experimental runs verify this assumption [8].
From (12.57) applied to the conduit (from forebay to surge tank) we have
(12.62)
where
T,, = elastic time for the conduit
hw = forebay head, per unit
504 Chapter 12

4 =zocqO
-- - normalized conduit impedance
h0
qc = conduit flow rate near surge tank, per unit
4Jc = friction coefficient for conduit
If we assume that the reservoir is large, we may write
hw= 0 (12.63)
since there will be no change in head at the forebay.
We now observe that, from Figure 12.15, that the per-unit flow rate at the surge-tank end of
the conduit is

4c = 4, + 4r + 4 p (12.64)
We can further describe the flow into the two tanks by the differential equation

Ttht = 41 + 4 r (12.65)
where T, = surge tank riser time.
Combining (12.62) and (12.64) and taking the Laplace transform with zero initial condi-
tions, we have

(12.66)

where

(12.67)

This equation is especially interesting since it indicates that the relationship between surge
tank head, h,, and penstock flow rate, qp,depends only on the conduit and surge-riser tank char-
acteristics and not on the characteristics of any component following the surge tank. In other
words, the hydraulic system up to the penstock is completely described by (12.66).

--
--

Tailrace

Fig. 12.15 Notation for changes in flow and head (all values are considered deviations from the quiescent values).
 Hydraulic Turbine Prime Movers 505

For the penstock, we apply equations (12.51) and (12.57) to write

h = (sech TeP)ht - zp(bh TePh - 4pq
1
qp = (cosh Tep)q+ -(sinh T,p)h (12.68)
ZP
where
qp = friction coefficient of penstock
Te= elastic time of penstock
zp=-
‘Oq0 = normalized impedance of penstock
h0

and all h’s and q’s are defined in Figure 12.15.

For the turbine, we may write the following equation [8]:
dq
q = -h + -n + -z34 = allh + aI2n+ a132 (12.69)
dh dn aZ
where
n = per-unit turbine speed
z = per-unit gate position

Also, we can write

dT,
T,= -h
dh
+ aT,
dn
n + -zdT,
dz
= aZlh+ aZ2n+ a2,z (12.70)

where T, is the per unit turbine mechanical driving torque. All values defined as a’s in (12.69)
and (12.70) are not constants but are nearly constant for any operating quiescent point. These
values will be read from curves of turbine characteristics.
Also from Newton’s Law, we have
dn
J,- = T, (12.71)
dt
where
J, = per-unit mechanical inertia
T, = turbine starting time
Here we assume no electrical torque as we are interested only in the relationship between
the variables, not in the way the turbine acceleration is restrained by shaft load.
Combining equations (12.63) and (12.65) we can write

(12.72)

where

(12.73)

which gives a relation between the per-unit turbine flow rate and the turbine head. We note that
it depends only on the characteristicsof the penstock, surge-riser tanks, and conduit, and not on
the turbine characteristicsas determined by partial derivatives in (12.63) and (12.64), nor on the
turbine inertia as given by (1 2.7 1).
506 Chapter 12

Hydraulic Water
.1-pp
Hydraulic
Supply Turbine System
(a) Hydraulic Components (b) Hydraulic System

Fig. 12. I6 Block diagrams of a hydraulic system.

Now, combining (12.69), (12.70), and (12.72) we get

(12.74)

Equation (12.74) is not yet in the desired form. Combining (12.69), (12.70), (12.72), and
(12.74), we can write

(12.75)

where

(12.76)

and

(12.77)

where
F6= "23 (12.78)
Finally, between (12.76) and (12.78) we deduce that

(12.79)

In block diagram notation, we can express the hydraulic system as shown in Figure 12.16.
Using equations (12.75) and (12.79), we have the representation of Figure 12.16 (a). We may,
however, lump these characteristics and use only (12.78) and Figure 12.16(b).

12.9 Simplifying Assumptions

It is quite apparent that the transfer functions (12.76), (12.77), and (12.78) are very diffi-
cult to work with and that some simplification would be helpful. One approach is suggested
at the end of Section 12.8. In this approach, a complex hyperbolic function is represented
by an infinite series and then higher-order terms can be deleted as an approximation. This
is a purely mathematical approach and is quite acceptable as long as the deleted terms are
small.
Another approach to simplification is through a combination of mathematical manipulation
and physical reasoning. This requires a certain amount of experience and intuition, and should
be verified by staged tests on a physical system.
Our approach is this latter method, drawing generously from the recorded thoughts of Old-
 Hydraulic Turbine Prime Movers 507

enburger and Donelson, as presented in [8]. These approximations are not only those devised by
experienced engineers, but tested extensively to prove their validity.
The first approximation noted is that concerning the hydraulic resistance. It is noted that,
although present in F,, F3, and all other factors (note +c and +p), the error in neglecting
the hydraulic resistance term is negligible. Thus, the resistance head-loss term we so careful-
ly added in equation (12.56) is not needed in the small-disturbance case. We will not bother
+
to remove the term in all expressions, but note that little error would result from doing
so.
One possible simplificationis that of neglecting the conduit portion of the hydraulic system
and assume that the surge tank isolates the conduit from the penstock. Thus, in equation (12.62)
we set the conduit flow to zero, i.e., +c = 0. This says that the water flow in the conduit does not
change and the conduit is essentially closed. Under this condition, from (12.64) and (12.65) we
have
qc = 0 = (41+ qr) + qp

Ttht = qr + qr = -qp

(12.80)

or
1
Fl = - (12.81)
Tts
and the surge tank acts as an integrator.
A second simplification involving F3 is possible from experience with physical systems.
We write

Fl
l+-hnhTp
1
ZP -
= (12.82)
F3(s)= +p + F , + Zp tanh T p Zp tanh T p
Both this assumption and the assumption on the isolation of the conduit (12.79) have been
validated by experiment.
We now examine certain approximations suggested by Oldenburger and Donelson [8],
which provide several degrees of simplification.

1. In the simplified expression for F3(s)from (12.82) we can set, as an approximation,

tanh T p 2 T p (12.83)
with the result

(12.84)

Using this approximation, we compute

(12.85)
508 Chapter 12

d2s2+ dIs + do
- (12.86)
e3s3+ e2s2+ e,s + eo

a2&i2s2+ dls + d0)(c2s2+ cIs + co) - az2(b1s+ b0)(e3s3+ e2s2+ els + eo)
+ a23(e3s3+ e2s2+ els + e0)(c2s2+ CIS + cg)
F6 =
(e3s3+ e2s2+ els+ eo)(c2s2 + c1s+ co)
-
-
5th Order Polynomial
(12.87)
5th Order Polynomial

2. Simplify F , by letting
1
F1= - (12.88)
ZCT2
and F3 by

(12.89)

and, finally, with

(12.90)

This results in a more complex model that is undoubtedly more accurate. In this case, the func-
tion F4is
5th degree polynomial
F4=
6th degree polynomial
and is much more detailed than the previous case. Experiments have indicated that, for all ex-
cept the most careful experiments, such detail is not necessary.

3. If the water in the conduit is assumed to be rigid, then equation (12.62) becomes [8]
h,-h,=Tc4c+4Ac (12.91)

In this case, F1 becomes a second order function:

T 2 + (6,
FI = (12.92)
TcTp2+ 4cTp + 1
and the other transfer functions also become higher order.

4. All of the above should be compared to the classical water-hammer formula based on a
lumped system:

(12.93)
 Hydraulic Turbine Prime Movers 509

Penstock Load Torque

Ref
Error
Signals
Servo
Stroke Position
Gate .1 Head
Turbine
Te
I
Shaft
Speed

Speed ~

Governor

Fig. 12.I7 Block diagram of a hydro turbine speed control system.

where Tw is the so-called “water starting time” (about one second). This gives a second-order
representation for F4.

In verifLing these approximations experimentally, Oldenburger and Donalson conclude

that the hydraulic system consisting of conduit, surge tank, riser tank, penstock, scroll case, and
draft tube can indeed be represented by a single transfer function relating Q to H a s in (1 1.71).
They verified that hydraulic resistance may be neglected without serious error. They note that a
second-order representation of F4 is adequate unless very accurate studies are to be performed.
The assumption that the surge tank isolates conduit and penstock systems is also verified.
Thus, although the hydraulic system is quite complicated, it may be represented adequately
for control purposes by a linear model in which all transfer functions are ratios of polynomials.

12.10 Block Diagram for a Hydro System

In considering the problem of controlling a hydro station, it is convenient to think of the
system block diagram, which is shown in Figure 12.17.
For a given steady load on the turbine T,, the electrical torque* is a constant and the speed
N will be that set by the speed reference p. This would be the case in an isolated system. In an
interconnected system, the speed is governed by the prevailing system frequency and the setting
of the reference p determines the load that will be assumed by this machine.
We can analyze the hydro system operation in a general way as follows. Any change in
speed is changed by the speed governor into a change in position or displacement x, which is
compared (usually mechanically) against a reference position p. Any difference in these posi-
tions produces an error signal cl, which is amplified by a control or servo amplifier to produce a
servo stoke Y, proportional to E, but having a much greater mechanical force to drive the wick-
et gates. This operation also usually introduces a delay or lag, which depends on the design of
the servomotor. The servomotor stroke Y repositions the wicket gates to produce a new gate po-
sition 2. In hydro turbines, the gate position is fed back mechanically as a means of adjusting
the droop or speed regulation.
In many hydro installations, the wicket gates are very large and massive. This means that
the servo amplifier must also be very large and capable of exerting large driving forces for mov-
ing such a large gate in a timely manner.

*It is common to represent the torque by the symbols Tor M.We use the There, but recognize that this symbol is also
used for time constants.
510 Chapter 12

12.1 1 Pumped Storage Hydro Systems

The hydro systems described above assume a storage reservoir of water that is elevated in a
configuration that will permit the water to be directed through a system of penstocks to hydro
turbines that are situated at a lower elevation. This is true of stations that use a storage system
fed by high-altitude streams, confined behind a dam. The confined water is held in storage until
power output from the station is needed, at which time it is used to power hydro turbine genera-
tors. This type of system is also used for a run-of-river system, where there is a continual flow
of water past the dam, some portion of which might be directed through hydro turbines to pro-
duce electric energy. In some cases, a minimum river flow might be necessary to support navi-
gation or other uses of the water downstream, even if the generators are unavailable for some
reason.
A pumped storage hydro power plant is different from the run-of-river system. In the
pumped-storage system there are two reservoirs, one at a high elevation into which water is
pumped for release later, usually at times of high system loading. This is accomplished using a
design of generator that can be operated efficiently as a motor and utilizing a turbine that can be
operated as a pump. There is a cost associated with providing the pumping power, which must
be performed at off-peak times when excess generation is available. Thus, there is an interesting
economic tradeoff between the cost of providing the pumped storage facility and the availabili-
ty of off-peak capacity to operate the pumps. Thus, the elevated water is not provided by nature,
but must be created by forcing the water into the elevated storage reservoir. If the pumping en-
ergy is available at a reasonable cost, and the generation provided by the pumped-storage plant
is of high value, then the overall economics of constructing such a facility may be quite attrac-
tive. The operating modes of a pumped storage system are shown in Figure 12.18.
Pumped storage plants require a suitable topology, where an elevated pool can be built
above the plant site. Aside from this physical restriction, there must be generation available for
pumping that can be obtained at a cost differential that will make the entire facility operation an
economic success. This requires the ability to pump power at a reasonably modest cost and a
higher energy value during the generating cycle. Such a variation of energy value on a daily ba-
sis is not uncommon, since peaking load usually requires the scheduling of peaking generation
with higher operating costs. Obviously, the economic parameters must be carefully evaluated in
considering the construction of a pumped-storage facility.

Fig. 12.18 The two operating modes of a pumped storage power plant.
 Hydraulic Turbine Prime Movers 51 1

Problems
12.1. Select a hydroelectric site of interest to you and record the physical features of the plant
including the type of turbine, the head, the installed capacity, etc. Document the sources
of your research and prepare a brief report on your findings.
12.2. Prepare a list of at least 10 hydroelectric sites, including a wide range of heads and phys-
ical features.
12.3. The system under study in [8] has the following constants:
Tec= 13 s f4,= 0.009 s
Te = 0.25 s 4, = 0.001 s
Jm=8s z, = z, =4

The base quantities are:

Torque: 40 MW at 225 rpm
Gate: 8 inches (at 80% of servomotor stroke)
Speed: 225rpm
Head: 428 feet (headwater-tailwater)
Flow rate: 1600 fi3/s
The turbine constants per unit are:
All = 0.57 A21 = 1.18
A21 = -0.13 A22= -0.35
A,, = 1.10 A23 = 1.5
Use approximation (1 2.70) and compute the following:
Fl ="us)
F3 =f,(F,,
tanh T2)
F4 =h(F3)

12.4. Find the transfer function of the hydraulic system shown in Figure 12.16 (b), where the
hydraulic supply and water turbine transfer functions are given by (12.75) and (12.79),
respectively.
12.5. Examine the effect of nonlinearity on the transfer functions F,, F3, F,, and F6 by using
the approximation
(a) tanh(Ts) = TS

(b) tanh( Ts)

(Ta3
= TS - -
3

(Ta3 2(TQ5
(c) tanh(Ts) = TS- -+ -
3 15
and finding the transfer functions for each F.
Use an approximating technique to factor the truncated polynomials of (a), (b), and (c)
and determine, by pole-zero plots, how the addition of extra terms in the series changes
the system response. Use the data from problem 3.
51 2 Chapter 12

References
1. Knowlton, A. E., Standard Handbookfor Electrical Engineers, Section 10, Prime Movers, McGraw-
Hill, New York, 1941.
2. Tietelbaum, P. D., Nuclear Energy and the US.Fuel Economy, 1955-1980, National Planning Asso-
ciation, Washington, D.C., 1964
3. Federal Power Commission, National Power Survey,1964, U.S.GovernmentPrinting Office, Wash-
ington, D.C., 1964
4. Notes on Hydraulic Turbines, Los Angeles Department of Water and Power, Private Communication.
5. Barrows, H. K., Water Power Engineering, McGraw-Hill, New York, 1943.
6. Craeger,W. P. and J. D. Justin, Hydroelectric Handbook, Wiley, New York, 1950.
7. Schleif, F. R., and A. B. Wilbor, The Coordinationof Hydraulic Turbine Governors for Power System
Operation,IEEE Trans. v. PAS-85, n. 7, p. 750-758, July 1966.
8. Oldenburger, R. and J. Donelson, “Dynamic response of a hydroelectricplant,” Trans. AZEE, Part ZZI,
81, pp. 403419, Oct. 1962.
9. deMello, F. P., Discussion of reference 8, Trans. AZEE, Part ZZZ, 81, pp. 418419, Oct. 1962.
10. Oldenburger,R. Frequency Response, Macmillan,New York, 1956.