Reverse Waterhammer

It has been a bias favoring the analysis of upstream hydraulic components including the
turbine, while less attention has been paid to the downstream water passage, often
treating the design of draft tube and tailrace as mere second thought. Yet, many important,
and potentially destructive, hydraulic phenomena occur in the draft tube, its extension,
and/or the pressurized tailrace tunnel, especially if the tailrace is long.

When deciding the closing speed for the guide vanes in a turbine a compromise between
the rotational speed of runner and pressure transients is necessary. If the closure is too
fast there will be severe pressure transients in the waterways. However, if the closure is
too slow the rotational speed will reach unacceptable levels. If the closure is too fast the
pressure in the draft tube can drop to under the vapour pressure of water. If the pressure
stays under the vapour pressure for an extended amount of time large vapour bubbles
called large cavities will form. These large cavities can grow large enough to cover a whole
section of the draft tube and form a full column separation. When the water comes back
the large cavities or the full column separation will collapse causing a large lifting force on
the turbine.
There have been a few severe incidents caused by a collapse of a column separation where
blades broke or where stations were lifted a few millimetres.

It is required to analyse available data from different closing tests to get a better
understanding of the hydraulic effects of a closure. These analyses will help create guidance
on how to find indications of large cavities when performing closure tests. If large cavities
can be detected at an early stage full column separation can be avoided.

Reverse Waterhammer in Kaplan Turbine

It is impossible to prevent the emergence of absolute vacuum conditions under the turbine
headcover (particularly in the case of high head Kaplan turbines, even if the closure time
is very long). 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. If measured data for the
pressure distribution at the turbine head cover (Hk) is not available, the mean pressure
drop through the runner can be calculated from the axial hydraulic thrust Fa. The condition
which prevents the runner from dewatering is Hk > Hvp + Hrez = 2 to 3 mWC, here Hvp is
the vapor pressure and Hrez takes into account pressure pulsations and calculation
accuracy. If this condition is not satisfied, the mean maximum axial outflow velocity, Cm
should meet the following criterion Cm <Ccr = 0.5 m/shich means that when the water
flow reverses from the draft tube and strikes against the rotating runner blades, provided
the water flow reverses at the same velocity with which it leaves the runner, the impact
force will be small and not cause damage. The values of Hrez and Ccr should be corrected
on the basis of practical experience gained in turbine operation.
Draft tube water column separation
In order to avoid the danger of water column separation and consequent reverse
waterhammer, the minimum pressure in any conduit (tunnel, penstock, pipeline), as a
design criterion, should not fall below 50 kPa (0.5 bar), even temporarily . This minimum
pressure must be the instantaneous value during a transient event and at the highest
position of waterway, typically at the runner outlet or at the upper limit of draft tube lining.
Since on site pressure transducers only can be attached to the wall of draft tube, and since
the water in the draft tube is usually in rotary motion, the measured pressure is actually
the maximum value for a given cross-section; the pressure in the vortex core, by contrast,
is likely at absolute vapor pressure or at the air release pressure of the dissolved air, or
slightly above this if air injected. Therefore, the measured pressures cannot be used
directly as a design criterion. Some manufacturers have empirical pressure and velocity
distribution data for their machines , and others have measured data from model turbine
draft tubes. All the data for each machine must be carefully evaluated because the
similarity laws do not apply to two-phase flows.

Analyses of draft tube water column separation

In practice, the lowest transient pressures are typically approximated using one-
dimensional waterhammer theory and transient modeling, which only represents the
“cross-sectional average” pressures which are usually higher than those found in the
vortex core and lower than those measured at the wall of draft tube cone. In addition, the
numerical analysis results can be uncertain and inaccurate around the unstable zones of
turbine characteristics, so the numerical analysis must be thoroughly performed and
experimentally confirmed. A larger safety margin is prudent for the minimum pressures in
the draft tube, particularly given the complexity and uncertainty of these transient flows,
as well the lack of an applicable similarity law. Moreover, the speed increment of the
runner, in runaway condition, is crucial since the voids formed by the centrifugal force of
the high-speed rotating water can be large; the pressure rise caused by even slow
accelerations/decelerations of the tailrace water may have a strong influence on void
collapse.

Vortex core and water column separation

For a hydro project it is particular important to understand the phenomenon of water
column separation and subsequent column rejoinder arising from the reverse water
hammer in the draft tube. Water column separation can occur during either steady or
transient conditions when low pressure establishes voids in the water column. Specifically,
this phenomenon occurs when the local pressure drops below the vapour pressure of
water or the partial pressure of dissolved gases (usually air). The dramatic pressure rise
that typically accompanies the collapse of such voids (water column rejoinder) can be
sufficient to crack the internal linings of conduits and damage both the turbine and other
hydraulic components. Perhaps more insidiously, this damage can progress unnoticed,
frequently observed only after repeated transient events, thus, creating a ‘time bomb’ of
future hydraulic and structural problems. However, the phenomenon of draft tube surge
has not been fully discussed in English although a few articles are listed in the References.
A predominant condition associated with reverse waterhammer is that the runner is
effectively dewatered when the flow reverses from the draft tube. Only if the axial
component of flow velocity is lower than the allowable value (around 0.5 m/s) will the
impact force on the runner blades from the reversing water be small and unlikely to cause
damage. In addition, runner lifting should also be checked during reverse waterhammer,
that is, the axial hydraulic force during the entire transient must be smaller than the weight
of the rotating parts. In horizontal machines thrust bearing controls this force.

In order to prevent severe rejoinder of partially separated columns of water, the large
vortex void in draft tube should be avoided. The pressure distribution in the draft tube
cone is inherently uneven (due to the helical motion of the flow), with the maximum
pressure occurring at the wall and the minimum near the center (Figure 1). To be more
precise, the minimum pressure occurs near the center of the vortex core, a complex
rotating tube of water, vapor and air, which usual forms in the draft tube. The vortex core
is unstable as a result of the swirl (tangential component of absolute velocity of water
outflowing from the turbine runner) induced by the action of the turbine blades. It is
important to recall that transducers mounted on the draft tube wall invariably record the
greatest cross-sectional pressures, instead of the lowest pressures in the vortex core at the
center of draft tube. In other words, even when the measured pressure is significantly high,
the core pressure could be at vapor pressure (or air release pressure), or perhaps slightly
above the vapor pressure if air has been injected into the system. Clearly, then, it is
inappropriate to rely on the pressures measured on the draft tube cone to judge whether
the minimum pressure exceeds 0.05 MPa. Computational fluid mechanics (CFD) and
numerical approaches clearly hold out some promise but they are unfortunately far from
being sound solutions to the challenge of measuring and predicting the possible range of
draft tube pressures and velocities due to the limited accuracy and numerical stability. The
difficulty arises from many sources, including the multi-phase, unsteady, asymmetrical,
three-dimensional and oscillatory nature of the flow, and the complex interactions
between the flow and the machine unit (see photos in Figure 1 and Figure 2).
Overall, the inherent complexity and significance of these events demand an empirical
approach to predict the occurrence of water column separation in draft tubes. It is
essential that all pressure data be carefully evaluated. This is particularly true since, we
emphasize, no similarity law is applicable to transient and/or two-phase flows while the
equations based on similarity still apply to the boundary conditions of the pump-turbine.
In practice, the lowest transient pressure in the turbine system is typically calculated based
on one-dimensional waterhammer theory and thus represents the “cross-sectional
average” value which is lower than the measured one but higher than that in the vortex
core, Figure 1.
Any mathematical solution in the zone of instability is highly inaccurate and uncertain;
fluctuations in the multiphase unsteady vortex core further complicate the flow conditions,
where model test data cannot be accurately transferred to the prototype.

Published history of reverse waterhammer

1.Sayano-Shushenskaya Catastrophe 2009
The official report, released on Oct. 3, 2009, blamed poor management and technical flaws
for the accident. According to the report, repairs on Turbine 2 were conducted from
January to March 2009, and a new automatic control system--meant to slow or speed up
the turbine to match output to fluctuations in power demand--was installed. [1] It seems
that this adjustment has been done without Adjustment or Modification in Project Design
or parameters have not been applied in commissioning and trial operation resulting in
catastrophic accidents and 76 deaths, 9 of 10 turbines were damaged or destroyed, and
6,400 MW supply, was lost.
Ref. [1]
[1] Sayano–Shushenskaya power station accident, (2009), Wikipedia,
http:// en.wikipedia.org/wiki/2009_Sayano%E2%80%93Shushenskaya_power_station_accident

concluded that reverse waterhammer (water column separation) was the source of
hydraulic force which as an explosion had ruined the turbines and generators. From Fluid
Mechanics and Hydraulics point of view only reverse water hammer could have been a
source of such force.

2.Stugun Power Station, 1992

An accident occurred in 1992 on commissioning testing at Stugun hydro power plant
located in the north of Sweden. Tests were made due to renovation of the generator and
the turbine. Reverse waterhammer has broken a runner blade and head cover. [2] A
Reconstruction, Redesign, Adjustment or Enlargement Design should have been reviewed.
Ref. [2]
Hillgren N., (2011), Analysis of hydraulic pressure transients in the waterways of hydropower stations, Teknisk-
naturvetenskaplig fakultet, UTH-enheten. UPTEC ES11007 Examensarbete 30 hp,
http://uu.diva-portal.org/smash/get/diva2:412851/FULLTEXT01

3.Akkats power station, 2002

In 2002, an incident occurred on commissioning testing at Akkats rebuild plant located in
Sweden. Although extensive repairs were made, it only proved to regain 80% of generation
capacity. During a load rejection the guide vanes were closed rapidly. A water column
separation collapsed and caused a lifting of the runner with 700 tons of shaft and generator
rotor [3]. A Reconstruction, Redesign, Adjustment or Enlargement Design should have
been again carefully accomplished
Ref. [3]
[3] Akkats Power Station , Vattenfall's European energy company,
http://powerplants.vattenfall.com/node/338

4.Kaplan turbine broken down in Sweden, 1986

Kaplan turbine has been broken down in Sweden in 1986. The turbine over speed signal was reached and emergency
stop level closed the guide vanes. The emergency closure at overspeed, a more severe initial condition, has been
enough to result in an accident.
Ref. [4]
[4]
Hillgren N., (2011), Analysis of hydraulic pressure transients in the waterways of hydropower stations, Teknisk-
naturvetenskaplig fakultet, UTH-enheten. UPTEC ES11007 Examensarbete 30 hp,
http://uu.diva-portal.org/smash/get/diva2:412851/FULLTEXT01
5.Ice Harbor Plant, 1965
Ice Harbor Plant, the US, Kaplan turbine No, 2 accident, 1962; unpublished, report.
Ref. [5]
[5] Report on investigation of accident to Turbine No.2 in PIT No.1 during sudden load rejection tests at Ice Harbor
Lock and Dam, (1962), Corps of Engineers, Walla Walla District, Contract DA-45-154-CIVENG-56-166 / S.O.
78930, February .2, 1962.

History of other cases

Cases of well-known plants such as Grand Coulee, Niagara Falls Plants, Iron Gate, Janpeg,
sometimes have no better choice or saved upfront money or in the worst case designed
by inexperienced experts or neglected design procedures (feasibility study, general design
detailed design, and reconstruction, particularly enlargement and/or increase of power) as
well as reviews and revisions by independent reviewers. Grand Coulee has no better choice
but to build in the side bank forming unfavourable “S” form inflow into the intake
structure, Figure 6; Sir Adam Back 2 Formed “S” inflow to save money upfront. Many
inexperienced designers copy this layout in spite that in these cases efficiency is reduced
and maintenance cost increased as in Site C project in Canada.

Sir Adam Back Plants

Sir Adam Beck plant turbines on the right side (Figure below) have unfavourable inflow of
water to save money at the construction. The inflow is not in the best hydraulic form.
Figure below is a very similar having similar inflow hydraulic conditions into the turbines
on the right side of the dam as previous Grand Coulee case (Figure at top). Saved
investments yet increased maintenance costs. Post factum comparison of investments,
maintenance costs and energy losses could give data for future analyses

Site C Project
Layout of the Site C Project under construction could have a better equipment layout, even
better than those with the previous two cases Grand Coulee and Sir Adam Beck.
Inflow of water into the turbines would be with less turbulence, and uneven pressure
pulsations which are a great challenge for designers and all the time for users. A very strong
reason must exist to make such layout. Even if computer calculations (CFD) and model tests
give good results, because of the lack of similarity with the prototype there might be a lot
of problems; vibrations, vortices, vortexes, and eddies will always be present!
Water hammer in hydropower plants is caused by turbine load acceptance and reduction,
load rejection under governor control, emergency shutdown and unwanted runaway, and
closure and opening of the safety shutoff valve. It induces pressure rise or drop in hydraulic
systems, rotational speed variation in hydraulic turbomachinery (pumps and water turbines)
and level fluctuation in surge tanks and air chambers.

Design principles of water hammer control strategies :

· (mitigation of excessive loads) including operational scenarios (closing and opening laws),
· surge control devices (flywheel, surge tank, regulating valve, air valve, etc.) and

· redesign of the pipeline components.

Most important, water hammer control means keeping transient pressure head and turbine
rotational speed rise within the prescribed limits.

The standard one-dimensional elastic column water hammer model is used for transient
analysis in case of long tunnel and penstocks.
An approximate analysis of long tailrace tunnel
The length of tailrace tunnel as a function of closing time of guide vanes and flow speed in
the tunnel in order to avoid the water column separation can be calculated applying the
approximate equation

where: L = permitted length of tailrace tunnel, Ts = closing time of guide vanes, k = 0.6 to
0.7 empirical coefficient, vdt = water speed at the runner outlet - the draft tube cone inlet,
hs = suction head vot = flow velocity in the tailrace tunnel. This approximate equation is
based on the rigid water theory neglecting the elasticity and as we know, is not published
in English.
In general design, detailed design, and reconstruction, particularly enlargement and/or
increase of power, modern computer method should be applied and result verified on site
in commission and trial operations.

The wicket gate can operate very fast like within 2 second to a complete close up.

Wicket Gate Changes for Full Load Acceptance

Time (s) - Wicket Gate Position (%)

0-100
1-50
2-0

This will help in frequency adjustment but you have to think about the over pressure created
in penstock. If cost of the reinforced penstock throws your project in the drain, you have the
reconsider the whole project.

A longer closure time for the wicket turbine can be feasible for Francis turbines if a
synchronous relief valve is installed.

Penstock Pressure Regulation with normal operation i.e. with load accepted or rejected
either slowly as the system requires or rapidly during faults, pressure water system follow
slow surge phenomena and depends upon the rate of closing the guide vanes. Wicket gate
closing time is always kept much greater than critical closure time (Tc) i.e. the time of
reflection of the pressure wave, this time,

Tc = 2l/ a

where l is the length of the pressure water system from tailrace to forebay/ surge tank and a
is the velocity of the sound in water (wave velocity).

APPENDIX
The Vibration of Turbines and Powerhouse
To produce low-cost hydroelectric power in dams, it is essential to focus on the
powerhouses. Hence, the key to finding the effect of turbines operating in the dam body is
to detect the hydraulic characteristics in the reaction turbines. The hydraulic characteristics
of Kaplan and Francis turbine units serve as the main engine of a powerhouse. Kaplan and
Francis turbines, which are classified as reaction turbines, are difficult to use under part-
load operation due to pressure oscillation. Solutions to the cavitation problem in draft tubes,
the vibration effect in powerhouses caused by operating the turbine will turn back as the
maximization of power generation, and the generation of low-cost power
Cavitation problem in the draft tube, vibration effect in the powerhouse due to the
turbine operation
Blade reshaping in a Francis turbine significantly improve the runner inlet flow performance
and cavitation.
Pressure distribution and frequency change with and without air admission, and the effect
of air admission decreases the low-frequency pressure pulsation and increases blade
frequency.
A low-pressure increase around the blades of the runner when the inlet valve is closed and
a velocity increase over the runner area.
Dynamic loads of turbines and generators are responsible for the creation of vibration in
hydropower, along with the natural causes like earthquakes and floods.
The vibration transmission caused by the hydraulic generators’ rotation is equivalent to
that of the vibrations caused by the fluctuation of the low-frequency tail .
Vortex trajectory expansion in the draft tube occurs due to a simultaneous increase in the
frequency and excitation.
In Francis turbines at low head , the cavitation rate, the magnitude of pressure pulsation, and
performance loss increase with increasing suction head
A structural joint filter and closed-cell foam help to reduce vibration transmissibility.
Improvements in the frequency and reduced amplitude of the vibration source is doable with
a better structural stiffness of the outer concrete structure of the spiral case.
Semi-cushion spiral structure is safe in terms of the the vibrational characteristics on the
spiral case structure