See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/309113974

Analysis of Rotor-Stator Interaction in Turbine Mode of a Pump-Turbine

Model

Article in Journal of Applied Fluid Mechanics · July 2016

DOI: 10.18869/acadpub.jafm.68.236.25086

CITATIONS READS

19 1,068

6 authors, including:

Deyou Li Hongjie Wang

Harbin Institute of Technology Harbin Institute of Technology
103 PUBLICATIONS 1,667 CITATIONS 104 PUBLICATIONS 1,589 CITATIONS

SEE PROFILE SEE PROFILE

Xianzhu Wei
Harbin Institute of Technology
90 PUBLICATIONS 1,567 CITATIONS

SEE PROFILE

All content following this page was uploaded by Deyou Li on 01 September 2017.

The user has requested enhancement of the downloaded file.

Journal of Applied Fluid Mechanics, Vol. 9, No. 5, pp. 2559-2568, 2016.
Available online at www.jafmonline.net, ISSN 1735-3572, EISSN 1735-3645.
DOI: 10.18869/acadpub.jafm.68.236.25086

Analysis of Rotor-Stator Interaction in Turbine Mode of a

Pump-Turbine Model
D. Y. Li1, R. Z. Gong1, H. J. Wang1†, X. Z. Wei1, 2, Z. S. Liu1 and D. Q. Qin1, 2
1 School of Energy Science and Engineering, Harbin Institute of Technology, Harbin, Heilongjiang, 150001,
China
2 State Key Laboratory of Hydro-Power Equipment, Harbin Institute of Large Electrical Machinery, Harbin,

Heilongjiang, 150040, China

†Corresponding Author Email: [email protected]

(Received May 18, 2015; accepted November 2, 2015)

ABSTRACT

The highest-level fluctuations in large pump-turbines are usually originated from rotor-stator interaction
(RSI) in the vaneless region. Hence, the studies of RSI phenomenon and corresponding unsteady effects are
significantly important to reduce the pressure fluctuations. In this paper, firstly, RSI in a pump-turbine,
featuring 20 stay vanes, 20 guide vanes and 9 runner blades, is analyzed through diameter mode theory,
which has been used widely. Then, 3-D unsteady numerical simulations are performed under six guide vane
openings in turbine mode. The comparison including performance and pressure characteristics between
numerical and experimental results shows a good agreement. Finally, best guide vane opening 21° is chosen
to analyze the distribution of pressure fluctuations. The detailed investigation of numerical results shows that
frequencies in the vaneless region at best guide vane opening are mainly blade passing frequency (BPF) and
its harmonic frequencies caused from RSI. The variation of BPF and its harmonic frequencies is confirmed by
diameter mode theory. For this type of the pump-turbine, the amplitude of 2BPF (18fn) shows the highest
corresponding diameter mode k2=-2, which indicates two high pressure regions caused by the component of
18fn in the vaneless region. Furthermore, the two high-pressure regions rotate in the counterclockwise
direction with rotational speed of the runner blades. This research could provide a basic understanding of RSI
to have a further study for pressure fluctuations in pump-turbines.

Keywords: Pump-turbine; Rotor-stator interaction; Turbine mode; Vaneless region; Diameter mode.

NOMENCLATURE

a opening of the guide vanes p time average calculation pressure

Bm amplitude of the mth harmonic T torque of the runner
Bn amplitude of the nth harmonic TB rotational period of the blade
D0 distribution diameter for guide vanes y+ non-dimensional wall distance
D1 runner inlet diameter Zg number of guide vanes
D2 runner outlet diameter Zr number of runner blades
f frequency Zs number of stay vanes
fn rotational frequency
H head of the pump-turbine Φm phase of the mth harmonic
m, n harmonic order Φn phase of the nth harmonic
Pr pressure field at runner inlet in rotating θr angle coordinate in rotating system
system θs angle coordinate in stationary system
Ps pressure field at guide vane outlet in w rotational speed
stationary system η hydraulic efficiency in turbine mode
p calculation pressure

1. INTRODUCTION lead to large fluctuations for the load of electrical

power network since they are strongly dependent on
New green energies such as wind and solar, have a meteorological conditions. In order to stabilize the
strong development during the last ten years, which electrical grid, a great number of hydropower plants
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

were developed due to their quick respond ability to In this paper, 3-D unsteady simulations of a pump-
the variation of the load (Decaix 2015), especially turbine, featuring 20 stay vanes, 20 guide vanes and
for pumped storage power plants. Because of 9 runner blades were conducted using the SST k-ω
effective storage and flexible switch between turbulence model in turbine mode. Based on the
turbine mode and pump mode, pumped storage validation of the experimental data including
power plants have a rapid development. Pump- performance and pressure characteristics, best guide
turbine as for the key component, the current trend vane opening 21° is selected to study the causes and
is the increasing of the power, head, and speed propagation characteristics of RSI combing
either for building new plants or for upgrading diameter mode theory in the vaneless region. In
existing plants. It leads to an obvious rise in the addition, the distribution of the pressure
pressure and fluid velocities (Rodriguez 2007). A characteristics in the circumferential direction and
great number of studies were carried out on flow vertical direction is also presented.
characteristics (Yin et al. 2010, 2014), instabilities
such as hump characteristics (Li et al. 2015) and S 2. PUMP-TURBINE SPECIFICATIONS
shaped characteristics (Wang et al. 2011), and
corresponding pressure fluctuations in pump- A reversible pump-turbine model is investigated in
turbines. Pressure fluctuation is one of hot topics in this research, which is reduced 8 times of a
pump-turbines. The increase of pressure prototype in a pump storage power plant (see Fig.
fluctuations leads to the fatigue problems more 1). Main parameters of the pump-turbine in turbine
common. Generally speaking, pressure fluctuations mode are listed in Table 1. The five components are
of pump-turbines in turbine mode operating at as follows: the spiral casing, the stay vanes, the
normal conditions for full load come from RSI. At guide vanes, the runner and draft tube. All the
part load they are mainly caused by the vortex rope measurements are carried out on the hydraulic
in the draft tube cone. At extreme off-design system in Harbin Institute of Large Electrical
operating condition, the pressure fluctuations are Machinery.
due to draft tube instabilities, runner channel
vortices and flow separation (Magnoli 2012). These
pressure fluctuations not only generate lot of noises
and serious oscillations, but also introduce
unfavorable characteristics. Hence, it is very
important to understand the causes of pressure
fluctuations to improve the overall performance and
reliability (Wang et al. 2001).
RSI has specific characteristics that could be clearly
observed in frequency and has been studied widely
through different methods. Tanaka (1990), Guo and
Marua (2005), Franke et al. (2005), and Guo and
Okamoto (2014) have used excited diameter mode
theory which was proposed by Kutota et al. (1983)
to determine vibration characteristics combing the Fig. 1. The test rig of the pump-turbine model.
number of stationary and rotating blades. These
studies could provide a quick and simple result. Table 1 Parameters of the pump-turbine model
Meanwhile, lot of researchers predicted pressure Parameter Value Unit
fluctuation caused by RSI with a reasonable D1 482.5 mm
accuracy using CFD. Wang et al. (2001) predicted D2 240 mm
pressure fluctuations in 2-D unsteady Zr 9 -
incompressible flow by using a vortex method, D0 563.5 mm
which shows a good agreement with the
Zg 20 -
experimental data. Yan (2010) obtained the pressure
Zs 20 -
fluctuations in the vaneless region using a
compressible CFD simulation and conducted that
the compressibility of water might strongly
influence the pressure fluctuation. Similar work was 3. EXPECTED ROTOR STATOR
carried out by Yin (2013) to study the effects of BEHAVIOR IN VANELESS REGION
water compressibility on pressure fluctuations.
More recently, a conclusion was obtained that the Pressure filed caused by RSI in the vaneless region
rotating of guide vanes has an obvious effect on the could be considered as a combination of inviscid
amplitudes of pressure fluctuations in the vaneless flow (potential) and viscous flow (wake). With
region (Li 2015). respect to potential effect, flow field in the vaneless
region is periodically perturbed by the rotating
Although a great deal of research has been carried runner blades. As for the viscous effect, the
out to investigate pressure fluctuations, a systematic hydrodynamic phenomena, which play a major role
investigation of RSI in the vaneless region using in RSI, is non-uniformity of the velocity field in the
numerical simulation, experimental validation and spiral casing, non-desirable flow angle in the
diameter mode theory is limited. The mechanism of distributor, flow separations and wakes (Zobeiri,
RSI is not fully clear, especially for pump-turbines. 2006).

2560
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.


p r  r ,t    B m cos  n  Z r  r  m  (2)
m 1

The final modulation can be expressed by Eq. (3) as

the product of both pressure fields. Fig. 3 illustrates
the modulation process between the rotating runner
blades and guide vanes.


p mn  ,t   A mn cos n  Z g  s  n  (3)
 cos  m  Z r   r  m 

Using trigonometric identities in Eq. (3) and

arranging the terms
Amn
pmn  s , t   cos  n  Z g   s  m  Z r   r  n  m 
2
A
Fig. 2. Computational domain.  mn cos  n  Z g   s  m  Z r   r  n  m 
2
(4)
According to the relation of stationary system and
rotating system, one can get the Eq. (5).
 r  s  t (5)

Eq. (4) can be expressed as Eq. (6).

a)
Amn
pmn s , t  
2

cos mZr  t   mZr  nZg s  n  m 
A

 mn cos mZr  t   mZr  nZg  s  n  m
2

(6)
Eq. (6) indicates the pressure fluctuation induced by
b) potential interaction in the stationary system, which
is the function of time and space. One defines two
diameter modes k1 and k2, which can be expressed
as Eq. (7) and Eq. (8).
k1  m  Z r  n  Z g (7)

k2  m Zr  n Zg (8)

Diameter mode k indicates the number of high-

c) pressure regions and low-pressure regions for a
Fig. 3. Modulation process between the runner frequency component in the circumferential
blade and guide vanes flow field a) flow field direction, for example, k=2 means that there are
distortion due to guide vanes wakes, b) flow field two high-pressure regions and two low-pressure
distortion due to rotating runner blades, c) regions for one frequency component. Since the
combination of both distortions (Zobeiri, 2006). pressure field will be influenced by the rotating
runner blades, the pressure distribution will change
with the time, which will rotate in the same
In the stationary system, the non-uniform flow field direction or opposite direction. The rotating speed is
at the guide vane outlet is caused by wake effect. shown as Eq. (9) and Eq. (10).
The loading of guide vanes creates a periodic flow
field (see Fig. 3-a). In the rotating system, periodic 1  m  Z r   k 1 (9)
flow field distortions are induced by the rotating
runner blades (see Fig. 3-b). These stationary and 2  m  Z r   k 2 (10)
rotating periodic flows can be expressed as the
following Fourier series according to diameter Sign symbol decides the rotating direction. When k
mode number. is positive, the direction is the same with the
rotational direction of the runner, vice versa. In this
 research, the pump-turbine features 9 runner blades
ps s ,t    B n cos  n  Z g  s  n  (1) and 20 guide vanes. The expected rotor stator
diametrical modes and corresponding to frequencies
n 1
of the fluctuations are summarized in Table 2.

2561
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

Table 2 Expected rotor stator diametrical modes

and corresponding to fluctuation frequencies
n m k1 k2 ω1/ω ω2/ω f/fn=mZr
1 1 -11 29 -0.8 0.3 9
1 2 -2 38 -9.0 0.5 18
1 3 7 47 3.9 0.6 27
1 4 16 56 2.3 0.6 36
a)
1 5 25 65 1.8 0.7 45
2 1 -31 49 -0.3 0.2 9
2 2 -22 58 -0.8 0.3 18
2 3 -13 67 -2.1 0.4 27
2 4 -4 76 -9.0 0.5 36
2 5 5 85 9.0 0.5 45
b)
According to other researchers’ studies, the
amplitude and power of pressure fluctuation is
usually considered much lower if the harmonic
order is higher. So only n=1, 2 are listed in Table 2.
In Kubota’s theory (1983), the absolute value of k
shows higher, the amplitude of corresponding
frequency is much lower, so k2 is usually neglected.
The absolute value k1 is 2, while n=1 and m=1,
corresponding frequency is 18fn whose amplitude
shows highest. In such case, the pressure field c)
rotates in 9fn in the opposite direction with the
runner. Then the k1 is -4, 5, 7, -11 and so on, the
corresponding harmonic frequencies are marked in
bold in the Table 1. It can be concluded that the
main frequencies in the vaneless region caused by
RSI are 9fn, 18fn, 27fn, 36fn, …, which are mainly
the blade passing frequency and its harmonic
frequencies. The amplitudes of their main
frequencies could be arranged by the value of k1.
Since the 36fn and 45fn, are relatively high, one
could neglect them in this analysis. As a
consequence, the amplitudes of these main d)
frequencies in a descending order are as following: Fig. 4. Meshes for different components a) spiral
18fn, 27fn, 9fn, 36fn, 45fn, … casing, b) draft tube, c) guide-stay vanes, d)
runner.
In a sum, the frequencies and their amplitudes in
the vaneless region are predicted through the
Table 3 Mesh details for different components
theory, which need to be validated through
numerical and experimental results in the Parameters
following parts. Component
Nodes y+

4. NUMERICAL SIMULATIONS Spiral casing 0.4 1488

Stay guide 0.7 1.82
4.1 Mesh Generation
Guide vanes 0.8 1.63
The commercial software ANSYS is employed in
this research to create the mesh and carry out the Runner 0.9 1.69
simulations. Structured hexahedral meshes for the
Draft tube 0.5 1408
whole computational domain are created using
ANSYS module ICEM. The computational domain Total (mil.) 3.4
is divided into five parts: spiral casing, stay vanes,
guide vanes, runner and draft tube (see Fig. 4). The
details of every components are summarized in
4.2 Boundary Conditions
Table 3. y+ for stay vanes, guide vanes and runner in The mass flow at spiral casing is used for the inlet
average is less than 2. boundary condition, which is set according to

2562
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

experimental data. The turbulence parameters are experimental and numerical results mainly come
specified in terms of turbulence intensity and from the following. Firstly, 13.4° guide vane
hydraulic diameter of the inlet. Static pressure (0 opening is small opening, which is an off-design
Pa) is set at the draft tube outlet (turbine mode). operation condition, in which the flow pattern is
Considering the recirculation and backflow, the extremely unstable and accompanies with flow
type of pressure outlet is opening. In addition, separation, vortex motion and secondary flow. So
smooth no-slip wall conditions are imposed for the the differences at this operation condition show the
rest of solid surfaces. Boundary function is adjusted highest. Secondly, leakage and gap losses as well as
to the low Reynolds number wall function. All the dis friction are neglected in the usual CFD set up.
cases are investigated for the energy characteristics Thirdly, the turbulence model based on URANS,
without considering the cavitation. which consists many empirical parameters, cannot
predict all kinds of flow. In this research, SST k-ε
4.3 Numerical Scheme turbulence model is chosen, and the differences for
torque, head and efficiency are less 5%, 3% and
All the simulations are conducted through ANSYS 2%, respectively, which satisfy numerical accuracy.
module CFX. 3-D incompressible unsteady A further analysis could be carried out based on the
Reynolds-averaged Naiver-stokes (URANS) validation above.
equations and mass conservation equations are
solved using finite volume method. Two-equation
turbulence model SST k-ω is chosen for closing
equations. In addition, the high resolution scheme is
used for the advection term and 1st order upwind
scheme is chosen for other terms. A time step
1.6×10-4s is set corresponding 1.2° of runner
rotation. Namely, a revolution for the runner needs
300 steps. The interface between rotor and stator is
Transient Rotor-Stator.

4.4 Validation for Performance

Characteristics
Six operation condition points of the pump-turbine
in turbine mode for the identical rotational speed
under different guide vane openings are chosen to
a)
validate the accuracy of the simulations. The
variation of the head, torque and efficiency with
guide vane opening is plotted in Fig. 5.

b)
Fig. 6. Monitoring points in the vaneless region
a) circumferential direction, b) vertical direction.
Fig. 5. Comparison for performance
characteristics between the experiments and
simulations. 4.5 Validation for Pressure Fluctuations
From Fig. 5, the difference between experimental Figure 6 shows the locations for monitoring points
and numerical results for all performance in the vaneless region. Fifty monitoring points are
characteristics curve is less than 5%. The maximal set in the vaneless region (between the runner inlet
differences for the head, torque and efficiency and guide vane outlet) close to the band (see Fig. 6-
between the experimental and numerical results are a) to investigate the pressure characteristics in the
observed at 13.4° guide vane opening, which are - circumferential direction. Five (A, B, C, D and E)
4.17%, -2.46% and -1.75%, respectively. The groups of monitoring points are set along the flow
minimal differences appear at 29° guide vane direction to investigate pressure characteristics in
opening. In a sum, the differences at large guide the flow direction. For each group, the distance of
vane opening show less than the ones at small guide every two monitoring points is equal. All the
vane opening, and the relative differences turn out monitoring points are equally copied four times in
much larger off the best guide vane opening (21°). the vertical direction (see Fig. 6-b) to obtain the
variation of pressure fluctuations in the vertical
The reasons for the differences between direction.

2563
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

The calculation results of last 2048 steps are used to The amplitudes close to the runner inlet show
study the pressure characteristics compared with obviously higher than the ones at the guide vane
experimental data. The frequency resolution is 3Hz. outlet, which indicates the effects of RSI are the
Dimensionless parameter ΔH/H is used to represent most serious close to the runner inlet.
how much the pressure fluctuation corresponding to
relative head. ΔH is peak-to-peak value during the
observing time. Point VL26 is an experimental and
also a numerical monitoring point in the vaneless
region. Fig. 7 gives the difference between the
experimental and numerical results for point VL26.

Fig. 8. Pressure fluctuation trend in the radial

direction.

Figure 9 illustrates the variation in the vertical

direction. The amplitudes close to the band (VL1-1)
and the hub (VL1-5) are a bit higher than the ones
in the other heights. The amplitudes in the different
Fig. 7. Comparison for pressure fluctuation of heights change little, the maximal difference is only
experimental and numerical results. 0.17% for group D. Hence, the monitoring points
near the band and close to the runner inlet are
From Fig. 7, it can be noticed that there is the same chosen to investigate the pressure fluctuation
trend between the experiments and simulations. The characteristics.
minimal difference 13.3% is observed at best guide
vane opening 21°. There appears a high difference
at off design operation condition, whose maximal
value reaches to 180.4%. In several studies
(Magnoli and Schilling 2012; Xiao et al. 2008,
2010; Liu et al. 2012), the differences of ΔH/H
between experiments and simulations also reach
around 200% at the off-design operation conditions.
The reasons for the differences mainly come from
the following. Firstly, the compressibility of water
and effects of fluid structure interaction (FSI) are
neglected in the simulations. Secondly, peak-to-
peak values might be not accuracy as the
experimental results, since the sampling time during
the simulations usually is less than the one for Fig. 9. Pressure fluctuation trend in the vertical
experiments due to the limited calculation resource direction.
and time.
The variation of ΔH/H in the circumferential
5. NUMERICAL ANALYSIS IN direction is shown in Fig. 10. Abscissa represents
the position at the circumference. The curve is
VANELESS REGION
generated through all the monitoring points in the
clockwise direction, which features obviously
The non-dimensional coefficient of pressure periodic. In addition, the distance between two
fluctuation is defined in Eq. (11), which stands for peaks (troughs) is 18°, which is the same as that of
how much the pressure fluctuations relative to the every two guide vanes. It indicates that pressure
head. In this research, the head is 50 meters. fluctuation appears a periodic variation passing
p p every guide vane. All the monitoring points are
Cp   100% (11) marked in the vorticity field (see Fig. 11). The plane
 gH
of vorticity field is chosen where all monitoring
21° best guide vane opening is chosen to simulate points are set. From the Fig. 11, it can be seen that
in order to investigate RSI without considering the points which show a peak value are all located
pressure fluctuations caused by other instabilities. in guide vane wake regions. Guide vane wake
The variation of ΔH/H for the five groups of regions usually consist wake space and momentum
monitoring points is shown in Fig. 8. It is found that exchange space, which show strong shearing action
the amplitudes of pressure fluctuations increase with the main flow and high turbulent fluctuations.
from the guide vane outlet to the runner inlet. For Composing the action of the rotating blades, the
group A, the amplitude is 2.7% at the guide vane amplitudes of ΔH/H in guide vane wake regions
outlet, which increases to 6.7% at the runner inlet. show rather higher than the ones in other regions.

2564
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

The wave trough points are located between the two field. When to t4=1.00TB, at this moment there
guide vanes. The fluctuations mainly come from appears another wave peak, so variation of pressure
potential interaction, so the amplitudes are fluctuation enters the next period.
relatively low.

Fig. 10. Variation in the circumferential a)

direction.

b)

Fig. 11. Vorticity distribution in vaneless region.

In order to get the pressure characteristics in the

circumferential direction, four points VL1, VL15,
VL26 and VL39, are chosen to be analyzed using
frequency spectrum and time domain methods. Fig.
12 shows pressure signals of different monitoring
points in one runner revolution. From every picture,
one can get the pressure fluctuation shows a
periodic variation for all the monitoring points, The
peak repeats every 5.28×10-3s (rotational period of c)
the blade). During this time, the runner rotates 40°.
It indicates that the pressure fluctuations are
obviously influenced by the rotating blades. Fig. 13
gives the pressure cloud chart of point VL1 at
different moments, corresponding with the time
marked in Fig. 12-a. Combining with contour of
pressure, the pressure for the point VL1 is mainly
influenced by pressure field caused by the head of
the runner blade. Since the existing attacking angle
in the head of the runner blades, one side will
appear high-pressure region because of attacking
while in another side there will appear low-pressure
region due to flow separation. When t1=0TB, the d)
pressure reaches to the peak, at this time, the Fig. 12. Pressure signals for different monitoring
monitoring point just locates in the high-pressure points a) VL1, b) VL15, c) VL26, d) VL39.
region. Then t2=0.18TB, the low-pressure region is
close to monitoring point, which leads the pressure Excluding point VL26, pressure signals of the rest
of the point to the lowest. As the runner continues to points have the same shape. The reason for this
rotate, at t3=0.52TB and t4=0.70TB, there are a small phenomena may be the result of different frequency
wave peak and a wave trough, respectively, which components integrating. In addition, the signal
come from RSI between the guide vanes and the phase shows very different among these four
runner blades. RSI generates non-uniform pressure monitoring points, which directly shows pressure

2565
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

a) b) c)

d) e)
Fig. 13. Contour of pressure for point VL1 at different time a) t1=0TB, b) t2=0.18TB, c) t3=0.52TB, d)
t3=0.70TB, e) t4=1.00TB.

a) b)

a) b)
Fig. 14. Frequency spectrum for different monitoring points a) VL1, b) VL15, c) VL26, d) VL39.

wave propagation in the circumference. between the guide vanes and the runner blades. For
point VL1, the first main frequency is 18fn, with an
Then, a Fourier analysis of all the pressure time amplitude of 1.73; the second main frequency is 9fn,
data is performed to determine the prevailing with an amplitude of 1.23. The amplitudes of 27fn,
frequencies at different locations (see Fig. 14). 36fn, and 45fn, are relatively small and show a
Where, f is real frequency, fn is rotational frequency, descending trend. This is the same with the results
20.87Hz; frequency resolution is 0.146fn. obtained through diameter mode theory.
It can be obtained that the main frequencies are However, the frequency characteristics of VL26
blade passing frequency and its harmonic show different from the other three monitoring
frequencies, such 9fn, 18fn, 27fn, 36fn, …. So the points. The amplitudes of 9fn and 18fn are much
frequencies in the vaneless region at best guide lower than those of other points, while the
vane opening are originated potential interaction

2566
D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

amplitudes of 27fn and 36fn show relatively high. It 0.15fn, fn, 4fn, 5fn. These frequencies are not
indicates pressure field of different frequencies produced by RSI and may come from other
components features different distribution and wave components, such as vortex motion, rope vortex in
propagation characteristics. the draft tube. A further study could be continued to
validate the sources of these frequency components.

6. CONCLUSION

A numerical investigation of a pump-turbine in

turbine mode is carried out combining with the
diameter mode theory. Numerical simulations are
performed at six guide vane openings under
identical rotational speed. Compared with the
experimental results, a good agreement is
demonstrated between them. The highest difference
for head, torque and efficiency is less than 5%, 3%
and 2%, respectively. With respect to pressure
fluctuation ΔH/H, there also shows the same trend
between the experiment and simulations. The
lowest difference is observed at the best guide vane
a) opening of 21°. Based on the validation of the
experiments, the variation of pressure fluctuations
in the vaneless region is analyzed. Main frequencies
are mainly blade passing frequency and its
harmonic frequencies. Pressure fluctuations in the
vaneless region at the best guide vane opening of a
pump-turbine in turbine mode, are mainly
originated from RSI between the guide vanes and
the runner blades. The first and second main
frequencies are 18fn and 9fn, respectively.
Moreover, the pressure distribution of 18fn rotates
in the counterclockwise direction with rotational
speed of the runner blade (9ω). Pressure
distribution and propagation characteristics for the
components of different frequencies are confirmed
b) through diameter modes theory. Based on the
Fig. 15. Pressure field distribution of 18fn at analysis above, a further study should be continued
different moment a) 0.1s, b) 0.10032s. to validate the sources of the low frequency
components. When an emphasis on the other
instabilities such as vortex rope, flow separation
In order to analyze the difference above, pressure and secondary flow is carried out, effects caused by
time data is filtered to get the component of 18fn. RSI could be filtered according to this study.
The time t1=0.1s and t2=0.10032s are chosen to get
the radar map (see Fig. 15). Namely, from the ACKNOWLEDGEMENTS
moment t1 to the moment t2, the calculation
continues 2 steps corresponding to 2.4° of the
This work was supported by Foundation for
runner rotation. Both pictures show the pressure
Innovative Research Groups of the National Natural
distribution exists two high-pressure regions and
Science Foundation of China (Grant No.
two low-pressure regions, which is the same with
51121004). The authors also would like to thank
the results predicted through diameter mode theory,
Harbin Institute of Large Electrical Machinery for
corresponding to k=-2 (18fn). In addition, the
its test platform for the experimental data of the
pressure distribution rotates in counterclockwise
pump-turbine.
direction with rotational speed of the runner blade
(9ω). It can be concluded that pressure fluctuations
in the vaneless region under the best guide vane REFERENCES
opening of a pump-turbine in turbine mode, are
mainly originated from RSI between the guide Berten, S., P. Dupont, M. Farhat and F. Avellan
vanes and the runner blades. Frequencies are mainly (2007). Rotor-stator interaction induced
blade passing frequency and its harmonic pressure fluctuations: CFD and hydroacoustic
frequencies. Moreover, pressure distribution and simulations in the stationary components of a
propagation characteristics for the components of multistage centrifugal pump, Proceedings of
different frequencies could be confirmed through FEDSM2007, 5th Joint ASME/JSME Fluids
diameter mode theory. Engineering Conference, San Diego, California
USA 37549, 963-970.
In addition, excluding the blade passing frequency
and its harmonic frequencies, there are still some Decaix, J., A. Mülle, F. Avellan and C. Münch
low frequencies with small amplitudes, such as (2015). Rans computations of a cavitating

2567
 D. Y. Li et al. / JAFM, Vol. 9, No. 5, pp. 2559-2568, 2016.

vortex rope at full load. 6th IAHR International stress of runners of very high head reversible
Meeting of the Workgroup on Cavitation and pump-turbines. International Journal of Fluid
Dynamic Problems in Hydraulic Machinery and Machinery and Systems 4, 289-306.
Systems, Ljubljana, Slovenia.
Wang, H. and H. Tsukamoto (2001). Fundamental
Franke, G., R. Fisher, C. Powell, U. Seidel and J. analysis on rotor-stator interaction in a diffuser
Koutnik (2005). On pressure mode shapes pump by vortex method. Journal of Fluids
arising from rotor stator interaction. Sound and Engineering 123, 737-747.
Vibration 39(3), 14-18.
Wang, L. Q., J. L. Yin, L. Jiao, D. Z. Wu and D. Q.
Guo, S. J. and H. Okamoto (2003). An experimental Qin. (2011). Numerical investigation in the
study on the fluid forces induced by rotor-stator “S” characteristics of a reduced pump turbine
interaction in a centrifugal pump. International model. SCIENCE CHINA Technological
Journal of Rotating Machinery 9(2), 135-144. Sciences 54, 1259-1266.
Guo, S. J. and Y. Maruta (2005). Experimental Xiao, Y., Z. Wang, Z. Yan, and J. Zhang (2010).
investigations on pressure fluctuations and Experimental and numerical analysis of
vibration of the impeller in a centrifugal pump pressure pulses characteristics in a Francis
with vaned diffusers. JSME International turbine with partial load. IOP Conference
Journal Series B-Fluids and Thermal Series-Earth and Environmental Science 12,
Engineering 48(1), 136-143. 012023.
Kubota, Y., T. Suzuki, H. Tomita, T. Nagafugi and Xiao, Y., Z. Wang, Z. Yan, Y. Luo, R. Xiao, G.
C. Okamura (1983). Vibration of rotating Peng and P. Xue (2009). Experimental and
bladed disc excited by stationary distributed numerical analysis of pressure pulsation in
forces. Bulletin of JSME 26, 1952-1957. Francis turbine, Proceedings of FEDSM2008,
2008 ASME Fluids Engineering Conference,
Li, D. Y., H. J. Wang, G. M. Xiang, R. Z. Gong, X. Jacksonville, Florida USA 55238, 1179-1185.
Z. Wei and Z. S. Liu (2015). Unsteady
simulation and analysis for hump characteristics Yan, J., J. Koutnik, U. Seidel and B. Huebner
of a pump turbine model. Renewable Energy 77, (2010). Compressible simulation of rotor-stator
32-42. interaction in pump-turbines. IOP Conference
Series: Earth and Environmental Science 12,
Li, D. Y., R. Z. Gong, H.J. Wang, G. M. Xiang, X. 012008.
Z. Wei and Z. S. Liu (2015). Dynamic analysis
on pressure fluctuation in vaneless region of a Yin, J. L., D. Z. Wang, D. K. Walters and X. Z Wei
pump turbine. SCIENCE CHINA Technological (2014). Investigation of the unstable flow
Sciences 58, 813-824. phenomenon in a pump turbine. SCIENCE
CHINA Physics, Mechanics and Astronomy 57,
Liu, J. T., S. H. Liu, Y. L. Wu, L. Jiao, L. Q. Wang 1119-1127.
and Y. K. Sun (2012). Numerical investigation
of the hump characteristic of a pump–turbine Yin, J. L., D. Z. Wang, L. Q. Wang, Y. L. Wu and
based on an improved cavitation model. X. Z. Wei (2013). Effects of water
Computer and Fluids 68, 105-111. compressibility on the pressure fluctuation
prediction in pump turbine. 26th IAHR
Magnoli, M. V., and R. Schilling (2012). Numerical Symposium on Hydraulic Machinery and
simulation of pressure pulsations in Francis Systems, Beijing, China, Tsinghua University.
Turbines. 26th IAHR Symposium on Hydraulic
Machinery and Systems, Beijing, China 15(6), Yin, J. L., J. T. Liu, L. Q. Wang, L. Jiao, D. Z. Wu
062029. Tsinghua University. and D. Q. Qin (2010). Performance prediction
and flow analysis in the vaned distributor of a
Rodriguez, C. G., E. Egusquiza and I. F. Santos pump turbine under low flow rate in pump
(2007). Frequencies in the vibration induced by mode. SCIENCE CHINA Technological
the rotor stator interaction in a centrifugal pump Sciences 53, 3302-3309.
turbine. Journal of Fluids Engineering 129,
1428-1435. Zobeiri, A., J. L. Kueny, M. Farhat and F. Avellan.
(2006) Pump-turbine rotor-stator interactions
Shi, F. and H. Tsukamoto (2001). Numerical study in generating mode: Pressure fluctuation in
of pressure fluctuations caused by impeller- distributor channel. Proceedings of 23th IAHR
diffuser interaction in a diffuser pump stage. Symposium on Hydraulic Machinery and
Journal of Fluids Engineering 123, 466-474. Systems, Yokohama, Japan 10, 1-10.
Tanaka, H. (2011). Vibration behavior and dynamic

2568

View publication stats