4.4 Waterhammer / Pressure Surges4.4.1 GeneralIn hydropower
installations, pressure surges occur as a result of changes in
velocity of flow in the penstock. Caused by either rapid closure or
opening of the control valve or by a sudden load rejection of the
turbine / PAT (failure of coupling, transmission belt, grid failure
in electricity generation), these pressure surges can reach several
times the static pressure of the installation. Precautionary
measures must be taken to avoid damage to pipelines and
machines.Waterhammer and pressure surges have been introduced in
Appendix A for the case of sudden and gradual valve closure. This
chapter deals mainly with load rejection of the PAT and the
subsequent pressure transients induced in the penstock.4.4.2
Gradual Closure/Opening for Nonlinear Valve CharacteristicsFigure
4.8 above has shown that valve characteristics are seldom linear
over the whole range of valve stroke. For many valve/penstock
configurations the rate of change of flow is low at the beginning
of valve closure and will increase rapidly towards complete
closure. The magnitude of possible pressure surges (waterhammer)
induced by valve closure will therefore depend on the maximum rate
of chance of flow rather than on the mean value. For practical
purposes an equivalent linear valve characteristic is defined as
shown in Figure4.10 below. The corresponding closure time should be
used for the waterhammer computations as shown in Appendix A for
gradual valve closure. Note that the valve opening(angle Delta )
might not always correspond to the valve stroke (revolutions of
wheel) because of the gate mechanism; a diagram of flow versus
stroke rather than versus opening should be employed to define the
equivalent valve closure time.Generally, a lever-actuated valve as
sometimes offered for small butterfly or ball valves should not be
used on PAT installations since very short closure times and
subsequently large pressure surges can be produced with this
mechanism.
FIGURE 4.10:Equivalent gate closure time for waterhammer
computations4.4.3 Pressure Transients due to Load Rejection of the
PATIn section 3.6.1, the problem of the runaway speed of a turbine
or PAT has been introduced.It was shown that the PAT is accelerated
to runaway speed when all external loads are removed. The values
given for runaway speed of different machines corresponded to the
final steady state conditions of the PAT. It was also shown that a
change of speed is in most cases accompanied by a change of flow
(except for turbines/PATs of nq ~ 80) which, as stated above,
basically bears the danger of inducing pressure surges into the
pipeline if the rate of change (DeltaQ/Delta t) is high. These
pressure surges or pressure transients might drive the PAT
temporarily to a higher speed than according to the steady state
runaway speed. Figure 4.11 below presents this transient behaviour
of a radial flow PAT and its interrelation with the penstock system
in a head-flow (H/Q) diagram (note the difference toFigure 3.31,
where the change of flow was assumed to take place very slowly
(along the system resistance curve) without inducing pressure
transients).
FIGURE 4.11:Transient pressure development and runaway speed
after total load rejection of a radial flow PATAs can be imagined,
the determination of the maximum PAT speed and pressure head after
full load rejection including waterhammer is rather involved.
Somewhat more analysis is required than for the computation of the
waterhammer due to valve closure/opening where a simple formula
could be given for a rough estimate (see Appendix A). While the
gate movement was determined externally by an operator, PAT
speed/flow variations after load rejection is itself affected by
the development of the waterhammer; this requires an iterative
solution of the problem.In the following, we will present a
simplified graphical method to estimate maximum waterhammer due to
load rejection of a PAT. Further development of this method would
lead to the general solution of transient problems in hydraulics
according to Schnyder / Bergeron.4.4.4 Machine ParametersWhen the
load on the PAT shaft is suddenly removed, the torque produced by
the water flowing through the impeller is no longer in equilibrium
with load torque. This torque difference therefore accelerates the
PAT until a new equilibrium between the driving hydraulic and the
resistive forces (friction) is reached. Figure 4.12 b) shows the
development of the torque from nominal to runaway speed under
constant head. As speed changes so does the flow through the PAT.
Figure 4.12 c) shows the speed versus flow relationship for a
radial flow PAT. Note that the flow through an axial flow machine
might be increasing with rising speed. Both curves T/n and Q/n are
developed from the PAT performance curve calculated in the course
of the PAT selection procedure discussed in Chapter 3. (Figure 4.12
a) To simplify calculations, a linear relationship between speed
and torque, and flow respectively has been assumed.
FIGURE 4.12: Development of torque and flow versus speed for a
radial flow PAT under constant headIt is now decisive for the
magnitude the waterhammer, in what time the PAT would accelerate
from the operating point (Qo) to runaway conditions (QR) The faster
the change of flow (Delta Q), the more powerful the induced
waterhammer will be.Acceleration of the PAT depends on the driving
torque and on the rotating masses (PAT impeller, pulleys, flywheel
and generator rotor or machine assemblies). Applying Newton's
second law of motion on rotating elements we will
find:ACCELERATION: dOmega / dt = T(Omega) / J (4. 1 ) where Omega =
rotational (angular) speed (rad/s)T (Omega) = accelerating torque
(Nm) as a function of rotational speed andJ = moment of inertia
(kgm) of all rotating masses (details of calculation see Appendix
F)As accelerating torque T(Omega ) we may use the speed - torque
relationship developed for constant head in Figure 4.12 a), knowing
that this will not be quite correct in many cases since head
increases due to pressure surges. Thus, acceleration may be
expressed with the following differential equation: dOmega/dt=T0/J
*(OmegaR-Omega)/( OmegaR-Omega0 ) ( 4. 2)Integrating this, leads to
an exponential law for rotational speed a: versus time. Figure 4.13
below indicates that the runaway speed OmegaR would be reached at
infinity. For our first estimate, we will use a straight line with
the slope of the curve at the starting point(Omega0). Thus, the
accelerating time becomes:Taeff=(OmegaR-Omega0)/Omega0 -Ta where (4
.3)Ta=J*Omega0/T0Ta is called the unit accelerating time.
FIGURE 4.13:Development of PAT speed versus time under a
linearly diminishing torque T4.4.5 Penstock ParametersThe main
parameter of the penstock is the reflexion time or periodTr = 2*L/a
where a = propagation velocity of the pressure wave along the
penstock and L = penstock length (see also Appendix A). It
represents the time required by the pressure wave to travel from
the PAT to the forebay and back. For compound penstocks (varying
diameters and/or pipe materials), use the equivalent wave velocity
and the equivalent pipe cross-sectional area defined as follows
(limits of application see below): a equi= L / (L1/a1 + L2/a2 +
L3/a3 ...+Ln/an) (4. 5) where L = total penstock length, a equi =
equivalent wave velocity, Li and ai are the length and wave
propagation velocities of the individual sections.A equi= L /
(L1/A1 + L2/A2 + L3/A3 ...+Ln/An) (4.6) where Aequi = equivalent
cross sectional area of the penstock and Li and Ai are the length
and cross sections of the individual pipe sections of different
diameter.4.4.6 Development of the Graphical MethodFrom the
waterhammer computations for valve closure/opening (see Appendix
A), we know that the reflection time Tr must be compared with the
closing time of the valve which, in the case of load rejection,
corresponds to the effective acceleration time of the PAT (formula
4.3). If acceleration of the PAT/generator unit is completed within
one period, i.e before the pressure wave arrives back at the PAT,
the full pressure rise according to Joukowsky's law
occurs:Taeff