Proceedings of the International Conference on CSHS03, Belgrade, 2003

REVERSE WATERHAMMER - CASE STUDIES

Prof. Dr A. GAJIû, Prof. Dr S. PEJOVIû, ass. B. IVLJANIN

Faculty of Mechanical Engineering, University of Belgrade,
27 Marta 80, Belgrade 11000
Invited paper
Abstract
Several cases and incidents caused by reverse waterhammer in hydraulic systems are
analyzed. The paper describes the fundamentals of mathematical models for analysis of
hydraulic transients in Kaplan turbines based upon the theory of the rigid waterhammer.
The case of the air introduction into the intake structure behind the inlet wheel gate is
described by a system of differential equations. The turbine behavior is accounted for by
model characteristics. The turbine governor is described by a system of the differential
equations. The computation of the hydraulic axial thrust also involves model test data,
while the runner blades thrust, pressure forces acting on the shaft and the runner hub are
analyzed separately.
Mathematical model and numerical program are verified by the field tests.
Computation results are compared with those obtained by measurement on site. A very
good agreement is received for all physical values analyzed, such as turbine head,
discharge, rotational speed, pressures in the casing, behind the runner, at head cover,
hydraulic axial thrust, etc. Interesting results concerning abnormal and catastrophic
transient regimes are also presented in the paper.
Key words: hydraulic, turbine, waterhamer, tests

1. Introduction

It is not always the case that the transient state pressure in the whole system is above the
vapor pressure of the liquid (water). In some low head systems or even in high head
systems in which transients are produced rapidly, the pressure in some part of it may be
reduced down to the vapor pressure of the liquid. This will cause the cavitation and may
produce vapor cavities in the flow or even column separation. If the system is not designed
to withdraw the high vacuum, the thin pipelines walls may be buckled, the pipe collapsing
inwards. Rejoining of the two separated columns or collapse of cavities results in reverse
waterhammer (RWH) with a large pressure rise. These pressure shocks may damage the
piping system, hydraulic machine, valve or/and any other part of the system.
If the pressure in a pipe drops down to the vapor pressure of the water (liquid), then the
cavity is formed and the water (liquid) column separation occurs. When the cavity is big
enough the column of liquid is divided into two separate parts. At the moment when the
cavity vanished, the two parts of water (liquid) columns rejoin. The theoretical pressure
rise due to contact of those two columns with different velocities may be calculated by the
Joukowski formula.
In cases of hydropower plants, pumping systems and water transportations the more
common title would be "Water column separation and reverse waterhammer".
The occurrence of a vacuum is usually disadvantageous, because it may cause strains,
which the system has not been designed to withstand. If the local pressure falls below the
vapor pressure then dissolved gases are released, water evaporates and a cavity is formed.

89
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

From that point the numerical analysis is based upon several hypothesis none of which
have been definitely proved to date.
In any long conduit (pipeline, tunnel, ...) reverse waterhammer may appear. First, a large
cavity is formed in the zone where the local pressure falls below the vapor pressure, and
later, when the cavity collapses, the separated water columns rush in, hitting each other.
Cases have been reported where this shock was strong enough to cause severe damages.
This is likely to happen in any plant with a long piping system. This event is very
dangerous because it may cause excessively high pressures, and even destroy the pipeline
or damage the machines.
The analysis should determine if the reverse waterhammer may occur or not. There is no
point in trying to design a system strong enough to withstand these shocks, therefore, the
design should be oriented towards preventing such an event. Two-phase flow and water
column separation is allowed only in the specially designed systems. Closed conduit flow
on the downstream side of the power plant could be very dangerous if the pipeline is too
long and water column separation occurs. Designers may decide to shorten the tailrace
tunnel in order to avoid a costly downstream surge tank, but the danger of too long
pressure tailrace is well known, and should be very carefully analyzed. Water column
separation and reverse waterhammer must be avoided in order to prevent costly accidents.
Even in cases of short pipelines reverse waterhammer occurrences were registered.
It is a well known fact that several accidents and damages had occurred in hydropower
plants (HPP) with Kaplan units due to reverse waterhammer (Krivchenko et al., 1975,
Pejovic et al. 1980, 1980-a), although the draft tubes were relatively short. Investigations
of the phenomenon were continued after some incidents already described in literature
(Pejovic et al., 1978; Pejovic et al., 1980-b, Gajic 1983). A numerical program has been
developed and verified by field tests (Gajic et all., 1992). The program consists of two
main parts; the first one is intended for the computation of the physical values of the unit
and the hydraulic system: discharges, pressures, runner speed, net head, etc., while second
part is intended for the computation of the hydraulic axial thrust and the estimation of the
occurrence of the reverse waterhammer. This program has been used for the estimation of
the reverse waterhammer occurrences in hydro powerplants with Kaplan and with bulb
units.

2. Reverse waterhammer in Kaplan turbines

The big accident had occurred in hydropower plant, rated data of which are: power output
22.4 MW, under a head of 19,3 m, with a discharge of 136 m /s, and a rotational speed of
150 r.p.m., runner diameter 4.65 m, see Fig. 2.1. A generating unit was operating during
the night with an output of 8 MW. Because of the governor failure, (the feedback tape was
torn off), the guide-vanes opened very quickly and the power output increased. The
instruments registered power oscillations between 20 and 25 MW, and roaring sound were
heard from the turbine room.
The unit was closed by pressing the turbine emergency shut-off bottom. The generator was
disconnected from the network. The unit overspeed control device went into operation.
When the machine came to stop, a banging noise was heard from the turbine and water was
leaking out through the turbine head cover. A serious accident had occurred.
One of the runner blades (No.2) was broken at the root, see Fig. 2.2.a. The fracture line
went obliquely from the hub towards the inlet edge. A smaller part of the blade root and
inlet section remained attached to the runner hub, while the larger part was broken off, see
Fig.2.2.b. The photo of the fracture, see Fig. 2.3., showed no cracks or defects in casting,
and no signs of material fatigue. It was clear that the break had started on the lower part of
the blade.

90
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Fig. 2.1 Cross-section through the HPP

Fig. 2.2.a Runner blade (No.2) Fig. 2.2.b Torn-off parts of the blade No.2
broken in the accident

Fig. 2.3. Appearance of fracture on Fig. 2.4. Runner blade No.4

The part of runner No.2 that damaged in collision with
remained on the runner the torn-off parts

91
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Damage, caused by collision with torn-off runner

blade was discovered at the inlet edges of all blades,
see Fig. 2.5
Damages were also found in the lower guide-vane
ring, upper guide-vane ring, runner pit liner cone and
inner head cover. The damage on the guide-vanes
were in the form of scratches, see Fig. 2.5. It can be
concluded, from the position and shape of score
marks, that the guide-vanes either had a very small
opening, or were completely closed at the moment
when broken blade struck them.After the accident
described, a series of tests were made with another
identical unit. An interesting strip chart, recorded
during the turbine's emergency shut-off, is shown in
Fig 2.5.
Fig 2.5. Scratches on the guide - vanes

Fig. 2.6. Quick closure from power output of 16 MW

The turbine closure time was 4 s. It is important to notice that the pressure on the turbine
head cover (h1 and h2 ) becomes an absolute vacuum at the end of the servomotor stroke.
It means that an air-vapor cavity is formed above the runner, resulting in water column
separation. Pressure on the turbine head cover begins to rise only 20 s after the beginning
or the transient process.
Much more information about the phenomenon and about the analysis of the reverse
waterhammer, the axial thrust, pressure distribution in the turbine and some other data
recorded on a model turbine under various steady operating conditions, are given by
Pejovic at all. 1978, 1980, 1980-a, Pejovic, Gajic 1988., Gajic 1993.,Gajiü at al., 1993,
2000, 2002.

2.1 WATER COLUMN SEPARATION AND VACUUM IN HYDROPOWER

PLANTS
2.1.1 Water column separation in penstock
A very typical design of hydropower plans system shows Fig. 2.7. If a negative pressure
occurs in the penstock, when the water level in the surge tank is low, than in the knee M at
the upper end of the penstock water column separation could happen. An accident caused
by such conditions occurred in an old Swiss power plant. A manhole located at point M
was damaged by the reverse waterhammer followed by previous water column separation.

92
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Water rushing from the open manhole flowed down to the power house, causing severe
damage (Jaeger, 1977).
In other similar cases, thin pipeline walls buckled when vacuum occurred, the pipe
collapsing inwards.

Fig.2.7. Pressure conditions causing vacuum in

a conduit at point M

2.1.2. Vacuum in tail race tunnel

Closed conduit flow on the downstream side of the power plant could be very dangerous if
the pipeline is too long and water column separation occurs. Designers may decide to
shorten the tailrace tunnel in order to avoid a costly downstream surge tank, but the danger
of too long pressure tailrace is well known, and should be very carefully analyzed. Water
column separation and reverse waterhammer must be avoided in order to prevent costly
accidents.

2.2. REVERSE WATERHAMMER IN THE LOW-HEAD TUNNEL

Fig. 2.8 presents the maximum and minimum pressure envelopes along the tunnel for
different cases of hydraulic transient in PS V2 both, computed and measured results ,and
shows that vacuum occurs in some parts of the system (Gajic et al., 1988).
Accident and damage of the pumps in the pumping station PS V1, see Fig. 2.8. occurred
during the transient when this pumping station was out of operation. Two centrifugal single
stage single suction pumps were out of operation and the valves on the pressure side were
closed. Volutes of both pumps were broken, see Fig. 2.9.

Fig. 2.8. Layout of the low pressure distribution system

93
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Fig. 2.9. Pressure envelops along the tunnel for different measured and calculated cases

Fig. 2.10. Broken volutes of the centrifugal pumps

3. Mathematical model
The mathematical model of the unsteady flow is based upon the theory of the rigid
waterhammer, while the case of air introduction into the intake behind the inlet wheel gate
is described by a system of differential equations. The turbine behavior is accounted for by
the model characteristics taking into account the added moment of inertia of the water
captured in the runner. The turbine governor is described by a system of the linearized
differential equations including all nonlinearities of various governor components.
The computation of the hydraulic axial thrust also involves model test data, while the
runner blades thrust, pressure forces acting on the shaft and the runner hub are analyzed
separately.

3.1 RIGID WATERHAMMER

Historically speaking, the development of waterhammer theory has followed a pattern of
increasing complexity. It started by using the solid body type analysis called "Rigid pipe -
incompressible fluid" theory. In that case the equations become :

94
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

dh 1 dc cc
 O 0 (1)
dx g dt 2D
The list of symbols is given in notations.
In this type of analysis it is assumed that pressure surges are not violent, therefore the
deformations of the pipes are negligible, and the water may be regarded as non-
compressible fluid. This is the case of Kaplan and bulb turbines, when the time for a
pressure and flow change is shorter then the pipeline reflection time 2 l/a. Reverse
waterhammer analysis in those cases are usually based upon the rigid waterhammer theory.

Fig. 3.1. Scheme of mathematical model Fig. 3.2. Intake structure

The scheme of a low-head HPP is given in Fig. 3.1. Unsteady flow in the system is
described by the following equations:
p a  pv 1
§ Ol · Q2
 g  z 0  z v   ¦ ¨¨  ] ¸¸ 1 2
dQ1 U 0 © 4 Rh ¹ 2A
1
(2)
dt dl
³0 A
pv  p a 7
§ Ol · Q2
 g z v  z 7  H   ¦ ¨¨  ] ¸¸ 22
dQ2 U 2 © 4 Rh ¹ 2A
7
(3)
dt dl
³2 A
In the case when discharges Q1 and Q2 are different and water elevation in the aeration
pipe z v is below z0 , then air flow appears into the inlet structure, see Fig. 3.2. The
corresponding equations are:
dVv
Q2  Q1 (4)
dt
dM v
Av U v c z (5)
dt
dz v Q2  Q1
(6)
dt Av
Mv
Uv (7)
Vv
pv U v RT (8)

95
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

2x pA
cz M
x 1 U A
1 1
§ 2 · x 1 § 2 · x 1
Uv UA¨ ¸ , if pv d p A ¨ ¸ (9)
© x 1¹ © x 1¹
N
ª º
2 x p A « § pv · N 1 »
cz M 1 ¨ ¸
x  1 U A « ¨© p A ¸¹ »
¬« ¼»
1 1
§ p ·x § 2 · x 1
Uv U A ¨¨ v ¸¸ , if pv t p A ¨ ¸ (10)
© pA ¹ © x 1¹
Mv 0 pv
Uv lim
Vv o0 Vv 0 RTV
(11)

Q2  Q1
cz (12)
Av
pv pA , Uv U A (13)
U Ac z cz
pv pA  , if c¢ 40 m/s (14)
2M 2
Rotational speed is determined by
dn 30 M h  M e
(15)
dt S J
The moment of inertia has to take into account all rotating parts of the unit Ju, as it was
described by Krivchenko at al. 1975, Pejovic at al. 1978, 1980, Duan 1983, including the
water mass captured in the runner Jad, see Fig.3.3.:
d w>cr @
Mh ³³³ U >cr @dV  ³³ U >cr @vn dA  ³³³ U dV (16)
dt V A V
wt
­ J u  J ad , if dn ¢ 0 ½
° dt °
° US 4 °
J ® J ad
32
De  d i l sr sin G ¾
4
(17)
° °
° J u , if dn t 0 °
¯ dt ¿

Fig. 3.3.Water masses in the turbine runner

96
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Turbine characteristics are given by the model tests propeller curves:

\ \ M , a0 , E  H
Mu e2 (18)
2g
P P M , a0, E  Mh
UDe Ae u e2
4
Mathematical representation of those curves is rather difficult, as it was described by
Boldy 1982., Gajic 1983., Martin 1987., 1990., Senoo Y., Yamaguchi M. 1986., 1987.

3.2 TURBINE GOVERNOR

The turbine governor is of the temporary droop type, where the feedback is taken from the
auxiliary servomotor. The block diagram of this type of governor is given in Fig. 3.4. The
transfer function for each component is listed in the block diagram. The constraints of
various governor components are: maximal and minimal positions of the auxiliary
servomotor, different maximum velocities of the piston movement in the guide-vane and
runner blade servomotors and maximal and minimal positions of these servomotors. The
head dependent relationship E k = f(y1), i.e. cam characteristics, are approximated by a set
of linear equations.

Fig. 3.4. Block diagram for a temporary droop governor

The differential equations with corresponding constraints for various components may be
written from the transfer functions listed in Fig. 3.4.:
- auxiliary servomotor
dy1  1
x  b p y1  e  c , y1min d y1 d y1max (19)
dt Ty1
- temporary droop feedback
de b p
c  x  e  b p y1   1 e, y Y
Ymax
(20)
dt T y1 d T
- guide-vane servomotor
dy 1
 y1  y , dy d dy , 0 d y d 1.0 (21)
dt Ty dt dt max
- runner blade servomotor
dE 1
E k  E , dE d dE , E min d E d E max (22)
dt Trk dt dt max
E k ay1  b, a0 a0 Y 

97
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

3.3 REVERSE WATERHAMMER ESTIMATION

To prevent occurrence of reverse waterhammer in Kaplan and bulb units it is necessary to
prevent dewatering of the runner during transient. The mean value of pressure on the
turbine head cover can be calculated on the basis of the measured axial thrust coefficient
on a model under various steady operating conditions, as it is described in literature
(Krivchenko 1975, Pejovic et al. 1978, 1980, 1980-a, Gajic et al. 1992,...). The pressure
distribution on the turbine cover can also be found by using local pressure coefficients
determined during model tests, (Krivchenko 1975, Pejovic et al. 1980-b, Liu
1987).
Data on axial thrust are especially important, since it is a frequent case that axial force
raises the runner, usually causing some damages.

3.3.1 Mean pressure above the runner - Integral method

To make a more complete analysis of the reverse waterhammer phenomenon, the axial
thrust was recorded during the model turbine tests under various steady operating
conditions. It is not always possible to prevent the appearance of absolute vacuum under
the turbine cover (in some cases, even if the turbine closure time is very long). Therefore,
an adopted criterion for reverse waterhammer states that the maximum pressure (usually
found near the guide-vanes) on the turbine head cover must, throughout the transient
process, be higher than the pressure of saturated water vapour increased by a reserve of
pressure that takes into account the approximate nature of computing procedures.
This condition, however, proves to be too severe, and an additional condition is introduced
for the case when the maximum pressure drops below a permissible value. The condition
in this case is that the mean velocity of water outflow under the runner should be lower
than a certain limiting small value. The axial thrust is usually expressed by the
nondimensional force coefficient V F , or by the unit axial force F11 :
Fa Fa
VV 4 2
, F11
K V UD n D2H

Fig. 3.5. Hydraulic thrust characteristic

98
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

The mean pressure drop 'H K through the runner can be calculated upon the axial
hydraulic thrust Fa :
Fa
'H K , (23)
> 
AR 1  rg2@
the mean value of pressure on the turbine cover will be:
ªc 2 º
H K H at  H S  'H S  K S « m »  'H K (24)
¬ 2g ¼
and the mean axial component of flow velocity at the runner outlet is:
cm Q (25)
 
ª D 2S 1  rg2 º
« »
¬« 4 ¼»
The condition which prevents the runner dewatering is:
H ² H vp  H lim
In case the above condition is not satisfied, the following additional criterion is
introduced:
c M ¢ clim
which means, that when the water reverses from the draft tube and strikes against the
runner blades, provided the water reverses at the same speed with which it leaves the
runner, the impact force will be relatively small and will not cause any damage. The values
of Hlim and clim must still be corrected on the basis of practical experience gained in
turbine operation.
The allowable meridional velocity and pressure are to be determined by analysis of
different types of turbines and turbine runners and draft tubes.
As the first approximation it could be considered:
clim = 0.5 m/s, H lim = 2 - 3 mWC
An additional condition is of equal importance. This is the time during which the runner
remains waterless, because both the size of cavity and the speed with which water
returning from the draft tube strikes at the runner depend on this. This condition is
specially suitable for analyzing oscillograms recorded when examining transient processes
in waterpower plants. If the duration of absolute vacuum is shorter than one second, it is
considered that there is no danger of accident because of waterhammer. These short
pressure drops are caused by pressure fluctuations and oscillations in the generating
unit.The dimensionless pressure on the ordinate is:
hm  H s
x
H
and abscissa is the relative radius r/R = d/D or relative diameter.
The model test results are use to determine the local pressure coefficient
h H
\ s 2g m 2 s
ue
on the turbine head cover (locations 01, 02, 03, 04, schematically given in Fig. 3.5.a) upon
discharge coefficient, v and head coefficient, j. Fig. 3.6 shows \ s 01 for the locations 01 at
runner blade inclination of E = 23 . The zero flow condition in the full closed positions of
guide vanes is rather complex in the runner chamber and in the draft tube of axial flow
turbines. Krivchenko, 1975. and Liu, 1987. presented more laboratory test results, with
pressure coefficients expressed in slightly simpler forms.

99
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Fig.3.6. Pressure coefficient for point 01

and for different ting points

For the chosen interval of time, and

from the waterhammer computa-
tions, coefficients of discharge M
and head \ are determined. From
the measured model test diagrams,
such as diagram shown in Fig. 3.6.,
pressure coefficient \ s is deter-
mined for the chosen location at the
turbine head cover. The absolute
pressure on the turbine cover can
now be determined by using the
expression:

u e2
Hs H at  'H s  H s  \s (26)
2g
The condition for the reverse water hammer, is that the runner is dewatered when the water
starts to flow back from the draft tube toward the runner. This will happen if, in the space
above the runner, the pressure is equal to that of the saturated vapor, provided that the
water flows out at meridional speed cm, higher than an allowable one, clim . Since the
computation is only an approximation which disregards pressure pulsation, a certain
pressure reserve H lim occurs in a sufficiently large zone above the runner, and mean value
of the meridional component of water outflow from the runner is
higher than the allowable speed clim .
It should be noted that the pressure distribution measured on the model, as well as the
equation (26), apply only when there is no absolute vacuum in the zone above the runner.
If that happens, the pressure distribution must be corrected, because the similarity
conditions are not fulfilled.

4. Numerical Analysis

4.1 COMPUTED RESULTS

Different transient regimes in HPP, shown in Fig. 2.1, including the case presented in Fig.
2.6, were analyzed after the accident described in chapter 2.2 (case II). Computed results
are presented in Fig. 4.1. The lines of guide vanes opening A, head H, discharge Q, and
speed n, are obtained by computation of the unit quick closure. To analyze the possibility
of RWH occurrence it is necessary to calculate, according to the methods already
described, the pressure distribution on the turbine head cover, or the mean pressure above
the runner, as was shown in Fig. 4.1.a. The time-dependent change of local pressures
H S 00 , H S 01 , H s 03 , and H s 04 , at the turbine head cover are given by broken lines. The
locations of the points inside the head cover are presented in Fig. 3.6.The line F shows the
hydraulic thrust variation. The mean pressure above the runner H , obtained from the
pressure change in the draft tube and the hydraulic thrust variation, is given by the full line.

100
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Fig.4.1. Reverse waterhammer calculation.

Closure time T=3.5 s

Fig. 4.1.shows that the mean pressure

drops below the allowable value at the
moment t=3.2 s from the beginning of
transient regime. According to the
adopted criterion water column
separation occurs. The water outflow
being higher than the allowable velo-
city:
c = 1.8 m/s > clim = 0.5 m/s,
which means that there is a danger of an
accident.
In the case of two-speed turbine closure,
with total closure time 14.5 s, the
pressure H drops below its limiting
value, but cm = 0.3 m/s < clim = 0.5 m/s
consequently, there is no danger of
accident (Pejovic at al. 1980-b). The
runner will be dewatered but the water returning from the draft tube will strike it with
small velocity and because of that with the small force too.

5. COMPUTED RESULTS - STUDY OF INCREASING POWER OUTPUT

One of analysed powerplants was "Iron Gate I" (HPP "Djerdap I"), during the study of
increasing power output from 172 to 192 MW. It has 12 Kaplan turbines with runner
diameter D = 9.5 m.
The cross section of the power Fig. 4.2. Cross section of the HPP
plant used for the verification of
the model is presented in Fig.
4.2.
Rated data are: discharge
3
800 m , power output 172 (later
s
increased to 194) MW, net head
27.16m, maximal 35.46 m,
minimal 17.5m., rotational speed
71.5rpm, runner diameter 9.5m.
Besides the fact that the
numerical program was already
tested and confirmed during the
several years use, it is necessary
to start every new analysis with a
verification of input data:
characteristic of the flow system, turbines, governors, gates and valves and other hydro
mechanical equipment. The best method for that verification is a comparison with the field
tests. The reasons for that are: use of turbine characteristics obtained by model tests in
stationary regimes, unsatisfactory reliability in small openings zone, insufficient data about
the plant, etc.

101
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

Results computed in a study described here were compared with those obtained by
measurement on site. The analyzes of the possibility of RWH occurrence were performed
according to the mean pressure method, because the model tests included axial hydraulic
thrust measurements, but not the measurements of the pressure distribution on the turbine
head cover. In spite of that, a very good agreement is received for all physical values
analyzed.
Load rejection from power output of 162 MW to the idle run is presented in Fig.5.1.
Experimental results are given by the solid lines, and computed results by dotted lines.
Notations are: Yy(mm)- guide vanes servomotor stroke, E ($) - runner blade inclination, n
(%) - runner speed , p sp (m) - spiral casing pressure, pdt (m)- draft tube pressure, Fu (kN) -
runner axial thrust. The turbine quick closure and stoppage are presented in Fig. 5.1. The
agreement of the experimental and computed results is obvious.

Fig.5.1. Load rejection and quick closure and turbine stoppage from 90 MW

6. Conclusions
The analysis of the various transient regimes can be done using the programs described in
the paper. For instance, manual stoppage from the runaway and the optimization of the
runner blade and guide vanes closure modes and other emergency cases which have not
been tested on site, and which were, consequently, analyzed mathematically. Study of
possible power output increase at the hydropower plant requires similar analysis for all
normal and emergency transient regimes which can occur in future operation.
Reverse waterhammer may be prevented by adjusting the time of turbine closure.
A sufficiently long closure time is the best protection. Governors with several closing
speeds, which permit slower closure mode at the end of closure, when the guide vanes
have small openings, are very suitable, since they permit a sufficient decrease in the unit's
rotational speed before the guide vanes are closed.

102
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

When control conditions require a minimal closure time, comprehensive and expensive
investigations must be conducted. The pressure distribution on the turbine head cover and
the hydraulic thrust should be measured on the model, under steady operating conditions. It
must be noted that in the case of cavitation, air introduction, or any kind of two phase flow,
there is no similarity between the model and prototype. The results should be critically
analyzed.
The guide vanes must be closed slowly, but the runner blades must be opened, or closed,
depending upon the initial steady conditions, if the governor fails to operate. If, at high
rotational speed and large runner blades inclination, the guide vanes close completely, the
axial thrust could raise the rotor, and serious accident may occur, because of the reverse
waterhammer.
When axial flow turbines operate in the brake or zero flow condition, and the guide vanes
are partially or fully closed, the axial force is negative. If it is greater then the weight of the
rotating parts, the rotor will be lifted.
The intensity of axial thrust is relevant to the type, size, suction head, speed of rotation and
inclination of the runner blades, as well as the aeration of the turbine. At the moment when
the turbine is completely closed (a = 0) the speed of rotation has passed its maximum, but
is still slightly smaller then its maximum value (n = nmax ). Therefore, the axial thrust has
reached its maximal intensity, which is highly dependent upon the runner blade inclination
and rotational speed.
In the further treatment of this problem, the limiting values of the maximum pressure on
the turbine head cover and the allowable values of the water outflow speed under the
runner should be thoroughly tested. Likewise, an analysis should be made of the influence
of quantities that are not modeled, such as air supply to the turbine, change in the
cavitation coefficient, and other quantities.

NOMENCLATURE
AR (m ) - runner sectional area
p (m) - draft tube pressure;
a0 (mm) - guide vane opening;
bt (-) - temporary speed droop;
c (-) - command signal;
cm (m/s) - mean meridional component of outflow velocity;
D (m) - pipe, or runner, diameter;
dg (m) - runner hub diameter;
e (-) - temporary droop feedback signal;
Fn(N) - runner axial thrust;
g ( m 2 ) - gravity acceleration;
s
h (m) - piezometric head;
H (m) - net head;
'H k (m) - pressure rise through the runner;
Me (Nm) - electrical torque;
n ( min 1 )- rotational speed;

Acknowledgements
The research project in which the investigations has been carried out is supported by
Ministry of science , technology and development Republic Serbia, through project 1328.

103
 Proceedings of the International Conference on CSHS03, Belgrade, 2003

References
Chaudhry, H.M., 1979, Applied Hydraulic Transients, VNR, New York.
Duan Chang Guo., 1985., Method and Program of Convergency for Computer Simulation of Hydraulic
Transients in Hydraulic Installation, Kexue Tongbao, Vol. 30., No.7., pp. 976-980.
Gajic A., 1983., Contribution to the Study of Nonstationary Phenomena in Hydroelectric Power Plants, (in
Serbian), Ph.D. Thesis, Faculty of Mechanical Engineering, Belgrade, pp. 115+310 Appendix.
Gajic A., Pejovic S., Obradovic D., Predic Z., 1988, Hydraulic Transients in the Low-Head Water
Distribution Tunnels, Proc. Int. Conference on Hydraulic Machinery, Turboinstitut, Ljubljana, pp.491-500
Gajic A., et al., 1990/1991., A Study of Normal and Emergency Transient Regimes of Kaplan Turbines at
HPP Djerdap I, (in Serbian), Faculty of Mechanical Engineering, Belgrade, Vol.1 - Vol.8.
Gajic A., Pejovic S., Arnautovic D., Ignjatovic B., 1992, Reverse Waterhammer Analysis in Kaplan
Turbines, Proc. 16-th IAHR Symposium Sao Paulo, Vol.1, pp. 161-172.
Gajic A., 1993., Kaplan Turbine Incidents due to Reverse Waterhammer and Mathematical Model Confirmed
by the Field Tests, Proc. Int. Symposium on Aerospace and Fluid Science, Sendai, pp. 741-766.
Gajic A., Krsmanovic Lj, Ignjatovic B., Predic Z., 2000., Transient Regimes of the Huge Kaplan Turbine
Measured in situ, Poc. 2-nd Int. Symposium on fluid machinery and fluid engineering, Beijing, Vol.1, pp.
480-486.
Gajic A., Ignjatovic B., Predic Z., 2002., Stresses of Kaplan Turbine Runner Blade During Transients, Proc.
21-st IAHR Symposium, Lausanne, Vol.2, pp. 887-894.
Guidelines for Analysis of Regulation of Hydroelectric Power Plants, 1977, Moscow, (in Rusian), Energia,
Moskow.
Huvet P.,1986., Influence de la Variation D'inclinaison des Pales sur les Regimes Transitories des Turbines
Kaplan, La Hoille Blanche, No 1, pp.138-147.
Jaeger C., 1977, Fluid Transients In Hydro:Electric Engineering Practice, Blackie, Glasgow
Krivchenko G.I., Arshenevsky N.N., Kvyatkovskaya E.V., Klabukov V.M., 1975, Hydraulic Transients in
Hydroelectric Power Plants, (in Russian), Energy, Moscow.
Liu Shang Zhi, 1987, Experimental Research on the Lifting-Rotor Characteristics of Axial-Flow Turbine,
Proc. 2-nd China-Japan Conf. on Fluid Machinery, pp. 239-248.
Martin C.S., Horlacher H.B., Correlation of Zero Flow, Locked rotor, and Runaway Characteristics of Pump-
Turbines, 1990., Proc. 15-th IAHR Symposium, Belgrade, pp. D5.1-D5.8
Obradovic D., Arnautovic D., Pejovic S., Gajic A., 1988., Mathematical Modeling of Transient Regimes in
Multi - Unit Hydro Power Plants, Proc. 14-th IAHR Symposium, Trondheim, p. C1., pp 163-175.
Pejovic S. Krsmanovic Lj., Gajic A.,et al. 1978, "A study of Reverse Waterhammer", (in Serbian),Faculty of
Mechanical Engineering, University of Belgrade, pp. 228.
Pejovic S., Krsmanovic Lj., Gajic A., Obradovic D., 1980., Kaplan Turbine Accidents and Reverse Water
Hammer, Third International Conference on Pressure Surges, Canterbury, pp. 391-399.
Pejovic S., Krsmanovic Lj., Gajic A., Obradovic D., 1980-a, Kaplan Turbine Incidents and Reverse Water
Hammer,Water Power and Dam Construction, August 1980., pp. 36-40.
Pejovic S., Gajic A., Obradovic D., 1980-b, .Reverse Water Hammer in Kaplan Turbines, Proc. X IAHR
Symposium, Tokyo, pp. 489-499.
Pejovic S., Boldy A.P., Obradovic D., 1987, Guidelines to Hydraulic Transients Analysis, Technical Press,
Aldershot
Pejovic S., Gajic A., 1989, Cases and Incidents Due to Hydraulic Transients - Yugoslav Experiences, Invited
paper et International Congress on Cases and Accidents in Fluid Systems, Sao Paulo Brazil, pp. 181-223.
Rabe J., 1985., Hydro Power - The Design, Use and Function of Hydromechanical, Hydraulic and Electric
Equipment, VDI, Verlag GmBH, Dusseldorf, pp. 684.
Time V. A., 1960, Reverse Waterhammer in Draft Tube of Kaplan Turbines, Electric Plants, No. 3, (in
Russian), Elektricheskie stancii, No. 3, 1960.
Wylie, B.E., Streeter, L.V., 1978, Fluid Transients, McGraw-Hill, New York
Zmud A. E., Latovskij Y. A., Robuk N. N., 1960, Reverse Waterhammer in Hydraulic Turbines, (in Russian)
Energomashinostroenie, No 2.

104