Professional Documents
Culture Documents
Kaplan Turbine Runner
Kaplan Turbine Runner
H O S T E D BY
Alexandria University
College of Mechanical and Electronic Engineering, Shandong University of Science and Technology, Qingdao, Shandong
Province 266590, PR China
KEYWORDS Abstract Different from other types of hydraulic turbines, the fluid–structure coupling vibration
Kaplan turbine runner; behavior of Kaplan turbine runners has still been studied limited before. One problem is that their
Internal blade-interactions; blades can rotate according to load changes, and the internal interactions among the blades via the
Acoustic fluid–structure flow field may produce an important influence on the added mass and added damping of the runner,
interaction; particularly when the blade angle is small. In this paper, the influence of internal blade-interactions
One-way fluid–structure on the added mass and added damping of a prototype Kaplan turbine runner has been studied
interaction; numerically. An isolated stage model from the end of the stay vanes to the bottom of the hub with
Fourier Transformation six blades was considered for simulation. The Acoustic Fluid-Structure Interaction (FSI) technol-
Method ogy based on the Finite Element Method was used to investigate the added mass effect first and
to provide the modal shapes and initial frequencies for the following one-way FSI analysis based
on the Finite Volume Method. The natural frequencies predicted by the Acoustic FSI were com-
pared with those from the one-way FSI analysis to validate the simulations. Then, the influence
of internal blade-interactions on the added mass and added damping, as well as the mechanisms,
were analyzed.
Ó 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria
University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).
Nomenclature
much more times one day than before, which leads to even lar- of hydraulic turbine runners in water is severely affected by the
ger forces. Such higher forces can produce higher vibration added mass and added damping effect from the surrounding
levels to the runners, which can cause fatigue damage and water, which is severely affected by the boundary conditions
shorten their lifetime. of the surrounding water domain[14,15]. Researches have
Kaplan turbines are one type of widely used hydraulic tur- shown that when a structure underwater moves closer to a
bines wich are mostly used in used in low water head and large rigid wall, the added mass and damping effect will increase sig-
capacity hydropower plants [3], and the blades of Kaplan tur- nificantly [14], and when it moves to a non-rigid wall, the
bine runners can rotate to make the runner operate under high increases of the added mass and damping effect are usually less
efficiencies for a wide range of operation. The typical structure significant than those of moving to a rigid wall, which depends
of Kaplan turbines is shown in Fig. 1. The excitation forces of on the stiffness of the non-rigid wall [15].
Kaplan turbine runners can be both static and dynamic pres- For Kaplan turbine runners, the boundary conditions are
sures [4,5]. The dynamic pressure mostly comes from the Rotor much simpler than those of Francis turbine runners and
Stator Interaction (RSI) [6-8]. Due to the aforementioned rea- pump-turbine runners, and the only nearby wall of the blades
sons, several failure cases for Kaplan turbine runners have is the chamber wall near the blade-tips. However, unlike the
been reported in the literature[9,10]. Resonances are also one Francis turbine runners and pump-turbine runners, the blades
important reason to cause fatigue damage because they can of the Kaplan turbine runners can rotate according to the load,
greatly increase the vibration level of runners [11]. To avoid and sometimes the neighboring blades can be very close and
resonances, studying the fluid–structure coupling vibration show a partially overlapped status, just as shown in Fig. 1.
behavior of Kaplan turbine runners is of great significance. The close neighboring blade is one special type of non-rigid
The fluid–structure coupling vibration behavior of Kaplan wall because all the blades have the same natural frequencies
turbine runners has been studied very little before. In another and mode shapes. When one blade gets closer to another
two papers by the author [12,13], the dynamic behavior of a one, its added mass and added damping may be affected signif-
single blade with and without a crack of a prototype Kaplan icantly due to its interaction with adjacent blades through the
turbine was studied experimentally, but the fluid–structure fluid field, thus the fluid–structure coupling vibration behavior
coupling vibration behavior of the runner underwater is still of the whole runner.
not available. The fluid–structure coupling vibration behavior The geometries of Kaplan turbine blades are similar to
those of rectangular plates. Kyeong-Hoon Jeonga [16], et al.,
used to study the vibration behavior of two fully overlapped
identical rectangular plates in water, and they found that each
mode of the rectangular plate separated into one in-phase
mode and one out-phase mode according to the vibration
direction of the two plates. The added mass of the in-phase
mode is always higher than that of the out-phase mode at each
distance, and the added mass of the in-phase mode decrease
with the distance increase, while the added mass of the out-
phase mode increases with the distance increase. However,
the added damping effect was not related, as well as there
are also no studies about the situation when two plates are par-
tially overlapped. The added mass and added damping change
with the Kaplan turbine blade rotation should be known to
avoid the resonances of the runner. The structure of Kaplan
turbine runners is similar to that of blade-disk structures.
Fig. 1 Close adjacent blades of Kaplan turbines. For blade-disk structures, each mode of the single blade will
2378 M. Zhang, Q.-g. Chen
form a mode family of the whole runner with different Nodal induced by the fluid, and x is the modal displacement of the
Diameter (ND) modes [17], and the Nodal diameters (ND) are structure. For one shape of the selected mode, uðx; y; zÞ, MS
formed by the points which remain stationary in the vibration and the FF ðtÞ can be calculated as:
cycle of a mode-shape. For different ND modes, the phase ZZZ
change between two adjacent blades is different, and due to MS ¼ qS u2 dV ð2Þ
the interactions among blades, the added mass and damping X
effect in water of different ND modes can probably be very dif- ZZ
ferent. For hydraulic turbines, the pressure variation of the FF ðtÞ ¼ sðtÞ udS ð3Þ
flow filed also has its Diameter and frequency due to the
C
rotor–stator interaction (RSI), which is determined by the
number combination of the stator vanes and blades [6]. Only where qS is the structure density, X the structure domain, s the
when both the Diameter and frequency are the same as those fluid surface load on the structure and C the fluid–structure
of one runner mode, this mode can be exited. To know the interface. The damping usually has little influence on the nat-
added mass and damping differences of different ND modes ural frequency, and the natural frequency of the system with-
can be helpful for avoiding resonances during operation and out consideration of the damping is:
for the safety design of the runner. KS þ KF
To experimentally determine the fluid–structure coupling x2n ¼ ð4Þ
MS þ MF
vibration behavior of hydraulic turbine runners is very difficult
because the runners are usually enclosed by the case and are and the damping ratio is:
difficult to be accessed. Therefore, numerical methods are CF
probably an ideal way. Due to the little influence of the water n¼ ð5Þ
2xn ðMS þ MF Þ
flowing on the added mass effect, it can be easily investigated
through an acoustic fluid–structure interaction (Acoustic FSI) where MF ; KF and CF are the added mass, added stiffness
technology [18,19]. However, due to the significant influence of added damping obtained from the fluid, respectively. The
fluid flowing on the added damping effect, its simulation is a added mass is the mass of fluid activated due to the vibration
little difficult, but there have been some numerical methods of the structure. The added stiffness describes the change in the
based on the one-way or two-way FSI simulation available flow-induced restoring force with the deflection of the struc-
in the literature [20-22]. Generally, though the methods based ture. The added damping represents energy extracted from
on the two-way FSI are more theoretically complete, the com- the structure as a result of work done by the fluid flow.
putation cost is usually very high, which makes it not suitable The parameters in Eq. (5) can be calculated by the flowing
for large submerged structures, like hydraulic turbine runners. simulations: modal analysis, steady-state CFD and transient
The biggest problem of the methods based on one-way FSI is CFD. For one selected mode, the modal analysis is mainly
that they need to project the modal shape from modal analysis used to get the natural frequency, modal mass and stiffness
to the blade profiles of the CFD flow domain, which makes the in vacuum and in water. The wet natural frequency and mode
modal shape change due to the water flowing not be able to be shape will act as the initial values for the following simulation.
considered. However, the mode shape changes due to water The steady-state CFD is mainly used to get the added stiffness
flowing for hydraulic turbine runners are usually not signifi- in water, and the transient CFD is mainly used to update the
cant. The computational cost of one-way FSI is much less than added mass and calculate the added damping in water. Some
that of two-way FSI, which makes it much more suitable for studies [22] have shown that the added stiffness of Kaplan tur-
the added damping simulation of hydraulic turbine runners. bine blades in water is only 1%~2% of the modal stiffness in
In this paper, the influence of internal blade-interactions on vacuum, which will provide little influence on the final added
the added mass and added damping of a prototype Kaplan mass and damping. Therefore, to save the calculation time,
turbine runner has been studied numerically. The acoustic the steady-state CFD step will be eliminated in this paper,
fluid–structure interaction (FSI) technology was used to inves- and the added stiffness in water will always be seen as 0.
tigate the added mass effect first and to provide the modal
shapes and initial frequencies. Then, they were applied to the (i) Modal analysis in vacuum and water
one-way FSI simulation shown in [22] to get the added mass
and added damping in flowing water. The added mass effect A modal analysis in vacuum provides the initial circular
of each mode from acoustic FSI and one-way FSI were com- natural frequency xv :
pared to validate the simulations. Finally, the influence and KS
mechanism of internal blade-interactions on the added mass x2v ¼ ð6Þ
MS
and added damping were analyzed.
The mode shape in vacuum can be extracted, and the modal
mass MS can be calculated from Eq. (2) or exported from the
2. Physical model and method
FEM software directly. The modal stiffness KS can be calcu-
lated from Eq. (6).
For a low damping structure in vacuum, the equation of Then, the wet modal analysis of the structure in water will
motion in water is [23]: be done, which can be achieved through the Acoustic FSI tech-
nology available in Ansys. The wet modal analysis will provide
MS x€ þ KS x ¼ FF ðtÞ ð1Þ
the frequency xf and mode shape in still water used in the fol-
where MS and KS are the structural modal mass and stiffness, lowing simulation. The added mass MF in still water can be
respectively. FF ðtÞ is the modal force on the whole structure calculated from the following equation:
Influence of internal blade-interactions on the added mass and added damping 2379
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 KS To purely investigate the influence of internal blade-
xf ¼ ð7Þ
2p MS þ MF interactions on the fluid–structure coupling vibration behavior
of runners, only six blades submerged in the water domain
were simulated. The blade model has been validated in another
paper by the author [12]. For simple, an isolated stage without
(ii) Transient CFD analysis considering the stay vanes were investigate, and the water
domain from the end of the stay vanes to the bottom of the
The harmonic modal motion of the structure is defined as hub was considered, which is also the method used in [24] to
xðtÞ ¼ X0 sin ðxn tÞ ð8Þ investigate the added damping of different ND modes of an
aero-engine turbine runner. This simplification can certainly
By projecting the motion to the flow field, a transient CFD
make the added damping of each ND mode in one mode fam-
simulation can be done. An averaged quantity, U, can be cal-
ily some different from that when the entire flow passage is
culated as:
considered. However, unlike the influence of the trailing-edge
Z 2pN
1 t0 þ xn shape of one hydrofoil on the added damping through affect-
/¼ FF xðtÞdt ð9Þ ing the Von-Kaman vortex sheeting in flowing water, the inter-
N t0
action mechanism of internal blade-interactions ought to have
where N is an integer number of oscillations and t0 an arbi- no relationships with the water flowing, which means that one
trary time coordinate. If the number of periods is high enough, certain ND mode with a higher added damping in still water
substituting Eq. (5) into Eq. (9) results in: ought to have a higher added damping even in flowing water.
Therefore, this simplification ought not to affect us to investi-
/ ¼ pX20 xn MF ð10Þ
gate the influence of Nodal Diameter on the added mass and
From Eqs. (9) and (10), the natural frequency can be added damping.
updated and used in the following transient CFD simulation. For the modal analysis in vacuum, only one blade was sim-
With some repetition, the convergent added mass can be ulated through Ansys Workbench19.2. The blade material is
obtained. The average modal work done by the flow on the stainless steel with Young’s modulus 1.93E + 11 Pa, a density
structure, W, can be obtained as: 7750 Kg/m3 and a Poisson’s ratio 0.31. The rod part of the
Z 2pN blade was removed, and the surface where the rod connects
1 t0 þ xn the blade profile was totally fixed. For the wet modal analysis
W¼ _
FF xdt ð11Þ
N t0 in water, the widely used Acoustic FSI technology was used to
simulate the added mass effect from the surrounding water. A
if the number of oscillations is high enough, W goes to its con-
fully damped solver with a fluid–structure coupled algorithm
vergent value:
was implemented. The acoustic body has a density of 1000
W ¼ CF pX20 xn ð12Þ Kg/m3 and a sound speed of 1483 m/s.
The wet modal analysis was divided into two steps: first,
From Eqs. (11) and (12), CF , can be obtained for the added
only one blade in water was simulated with all other blades
damping ratio calculation in Eq. (5).
suppressed, and then, all the blades were simulated together.
The surfaces that the simulated blades connect the water
3. Simulation setup domain was set to be fluid–structure interfaces, the inlet and
outlet surface were set to be totally absorbed surfaces (the
The investigated turbine is the same one described in [12,13], absorption coefficient is 1), and all other surfaces were set to
which is located in Spain. The turbine has a head of 34 m be rigid walls (including the profiles of the suppressed blades).
and a maximum power of 73 MW. The runner has 6 blades Totally three blade-angles were investigated: 20-deg (the angle
rotating at 125 rpm, and the distributor has 24 guide vanes. nearly at the best efficiency point), 10-deg and 0-deg (totally
The flow rate at the best efficiency point is about 22.5 m3/s. closed), which has been shown in Fig. 2. A tetrahedral mesh
The diameter of the runner is about 6 m, and the tip clearance
is 0.09% the outlet diameter of the runner. Viewed from the
stationary frame, the RSI frequency depends on the rotating
speed of the runner (ff ), the number of rotating blades (Zb )
and the order of harmonics (n):
fb;n ¼ n Zb ff ð13Þ
The main excitation shape (k, the diameter of one excita-
tion mode) that can be calculated as:
k ¼ m Zv n Zb ð14Þ
where Zv is the number of guide vanes, Zb the number of the
blades and m the order of harmonics from the rotating view.
The highest amplitudes are obtained for the minimum values
of |k|. For the turbine studied, the minimizes |k| is obtained
from m = 1 and n = 4 with k = 0. For m = 1, n = 4 and
k = 0, the nominal speed 125 rpm corresponds to an RSI fre-
quency 50 Hz. Fig. 2 Blade angles investigated.
2380 M. Zhang, Q.-g. Chen
Fig. 3 View of the mesh a) Mesh for the Acoustic FSI b) Mesh Fig. 5 Mode shapes of the single blade model in vacuum and in
for the one-way FSI. water (0-deg).
Influence of internal blade-interactions on the added mass and added damping 2381
mented, and this was also applied to all other cases in this
paper. As can be seen, the projection of the 0 ND mode shape
in water yields a frequency 0.86% higher than that of the single
blade in vacuum, which demonstrates that the mode shape
changes from vacuum to water and the differences among dif-
ferent ND modes have little influence on the fluid–structure
coupling vibration behavior. The mode shape of the single
blade in vacuum under the corresponding blade angle was used
for all the cases of different ND Bending modes under differ-
ent blade angles in the rest parts.
All the simulated natural frequencies from the Acoustic FSI
and the one-way FSI for different ND modes in the Bending
mode family have been shown in Table 2. The errors between
the Acoustic FSI and One-way FSI are less than 3%, even
Fig. 6 Mode shapes of different ND modes in the Bending mode though the influence of the added stiffness is not considered.
family in water (0-deg). Under each blade angle, the one-way FSI yields a frequency
sequence of different ND modes approximately the same as
that of Acoustic FSI. Overall, the limited errors between the
Acoustic FSI based on the Finite Element Method and the
One-way FSI based on the Finite Volume Method validate
our simulations.
The frequencies of BD modes are a little lower than the RSI
frequency, which determines that they are more prone to be
exited during the star-up stage or load variations. Due to the
high computational cost of one-ways FSI simulation, it was
only done for the different ND modes of the BD mode family
under different blade angles. In the following parts, the natural
frequency and added mass analysis will be conducted on the
results of Acoustic FSI, which can show the results of different
mode families. The added damping analysis will be conducted
Fig. 7 Modal forces.
on the results of one-way FSI for the BD mode family.
Table 2 Frequency comparison between the results of Acoustic FSI and One-way FSI.
0 ND 1 ND 2 ND 3 ND
0 deg Acoustic FSI (Hz) 33.51 39.69 41.87 42.23
One-way FSI (Hz) 34.41 40.87 42.79 42.91
Error/% 2.62 2.89 2.15 1.58
10 deg Acoustic FSI (Hz) 37.30 40.81 41.84 41.79
One-way FSI (Hz) 37.95 41.44 42.91 42.73
Error/% 1.71 1.52 2.26 2.19
20 deg Acoustic FSI (Hz) 40.03 42.07 41.94 41.53
One-way FSI (Hz) 40.52 43.04 42.73 42.38
Error/% 0.97 2.31 2.13 2.05
2382 M. Zhang, Q.-g. Chen
Fig. 8 Normalized added masses of different ND modes under different blade angles for a) Bending b) T (0,1) c) T (0,2) d) T (1,2).
Fig. 9 Normalized natural frequencies of different ND modes under different blade angles for a) Bending b) TS (0,1) c) TS (0,2) d) TS
(1,2).
Influence of internal blade-interactions on the added mass and added damping 2383
be higher than 30% for different BD ND modes with an added [9] C.V. Frunzaverde D, D. Nedelcu, et al., Failure analysis of a
mass difference more than 100% when the blade angle is low, Kaplan turbine runner blade by metallographic and numerical
but the differences are limited for all the mode families when methods, in: Proceedings of the 7th WSEAS International
the blades are totally open. The damping ratio of the 0 ND Conference on FLUID MECHANICS (FLUIDS’10),
University of Cambridge, UK., 2010, p. 60–67.
Bending mode always is highest under each blade angle, which
[10] Y. Luo et al, Fatigue of piston rod caused by unsteady,
is 19% under the 20-deg blade angle and 112% under the 0-deg unbalanced, unsynchronized blade torques in a Kaplan turbine,
blade angle higher than the lowest damping ratio in the bend- Eng. Fail. Anal. 17 (1) (2010) 192–199.
ing mode family under each blade angle. [11] X. Liu, Y. Luo, Z. Wang, A review on fatigue damage
The influence of blade-interactions on the added mass is the mechanism in hydro turbines, Renew. Sustain. Energy Rev. 54
synthetic effect of many factors, like the global disturbance on (2016) 1–14.
the flow field from all the blades’ vibration and the local distur- [12] M. Zhang et al, Failure investigation of a Kaplan turbine blade,
bance from the vibration phases of adjacent blades in the over- Eng. Fail. Anal. 97 (2019) 690–700.
lapped area, etc. In this research, the influence of the guide [13] M. Zhang et al, Experimental and numerical investigation on the
vanes is not considered. For real Kaplan turbines, when the influence of a large crack on the modal behaviour of a Kaplan
turbine blade, Eng. Fail. Anal. (2020) 104389.
blades are totally closed, the guide vanes may also close. A sig-
[14] D. Valentı́n et al, Experimental study on the added mass and
nificant potential influence of the guide vane status has been damping of a disk submerged in a partially fluid-filled tank with
shown. However, whether the closed guide vanes can be seen small radial confinement, J. Fluids Struct. 50 (2014) 1–17.
as rigid walls and how much impact does it have still need [15] D. Valentı́n et al, Experimental study of a vibrating disk
experimental validations in the future. submerged in a fluid-filled tank and confined with a nonrigid
cover, J. Vib. Acoust. 139 (2) (2017).
Declaration of Competing Interest [16] K.-H. Jeong, G.-H. Yoo, S.-C. Lee, Hydroelastic vibration of
two identical rectangular plates, J. Sound Vib. 272 (3–5) (2004)
The authors declare that they have no known competing 539–555.
[17] M.P. Castanier, C. Pierre, Modeling and analysis of mistuned
financial interests or personal relationships that could have
bladed disk vibration: current status and emerging directions, J.
appeared to influence the work reported in this paper. Propul. Power 22 (2) (2006) 384–396.
[18] C.G. Rodriguez et al, Experimental investigation of added mass
Acknowledgments effects on a Francis turbine runner in still water, J. Fluids Struct.
22 (5) (2006) 699–712.
The authors wish to acknowledge the economic support [19] E. Egusquiza, et al. Fluid added mass effect in the modal
received from the Natural Science Foundation of Shandong response of a pump-turbine impeller, in: ASME 2009
Province (No: ZR2018MEE036) and Project of Shandong Pro- international design engineering technical conferences and
vince Higher Educational Young Innovative Talent Introduc- computers and information in engineering conference,
American Society of Mechanical Engineers, 2009.
tion and Cultivation Team [Performance enhancement of deep
[20] Y.S. Zeng et al, Numerical investigation into the effect of the
coal mining equipment]. trailing edge shape on added mass and hydrodynamic damping
for a hydrofoil, J. Fluids Struct. 88 (2019) 167–184.
References [21] B. Nennemann, C. Monette, J. Chamberland-Lauzon,
Hydrodynamic damping and stiffness prediction in Francis
[1] C. Valero et al, Extension of operating range in pump-turbines. turbine runners using CFD, IOP Conference Series: Earth and
Influence of head and load, Energies 10 (12) (2017). Environmental Science, IOP Publishing, 2016.
[2] I.E.A. IEA, Key electricity trends 2017 based on monthly data, [22] J.P. Gauthier et al, A numerical method for the determination of
Monthly Electricity Statistics, 2018. flow-induced damping in hydroelectric turbines, J. Fluids Struct.
[3] H.R. Krishna, Hydraulic design of hydraulic machinery, 69 (2017) 341–354.
Avebury, 1997. [23] D. Woyjak, Acoustic and fluid structure interaction, a revision
[4] S. Liu, S. Li, Y. Wu, Pressure fluctuation prediction of a model 5.0 tutorial. Swanson Analysis Systems. Inc., Houston, 1992.
Kaplan turbine by unsteady turbulent flow simulation, J. Fluids [24] H.D. Li, L. He, Blade aerodynamic damping variation with
Eng. 131 (10) (2009) 101102. rotor-stator gap: a computational study using single-passage
[5] L. Zhou et al, Analysis of dynamic stresses in Kaplan turbine approach, J. Turbomach. 127 (3) (2005) 573–579.
blades, Eng. Comput. 24 (8) (2007) 753–762. [25] S. Connell, et al. A comparison of advanced numerical
[6] C. Rodriguez, E. Egusquiza, I. Santos, Frequencies in the techniques to model transient flow in turbomachinery blade
vibration induced by the rotor stator interaction in a centrifugal rows, in: Turbo Expo: Power for Land, Sea, and Air, 2011.
pump turbine, J. Fluids Eng. 129 (11) (2007) 1428–1435. [26] C. Monette et al, Hydro-dynamic damping theory in flowing
[7] R. Iovănel, D.-M. Bucur, M. Cervantes, Study on the accuracy water, IOP Conference Series: Earth and Environmental
of RANS modelling of the turbulent flow developed in a Kaplan Science, IOP Publishing, 2014.
turbine operated at BEP. Part 1-velocity field, J. Appl. Fluid [27] E. Askari, K.-H. Jeong, M. Amabili, Hydroelastic vibration of
Mech. 12 (5) (2019) 1449–1461. circular plates immersed in a liquid-filled container with free
[8] R.G. Iovanel, G. Dunca, M. Cervantes, Study on the accuracy surface, J. Sound Vib. 332 (12) (2013) 3064–3085.
of RANS modelling of the turbulent flow developed in a Kaplan [28] A. Presas et al, Influence of the rotation on the natural
turbine operated at BEP. Part 2-pressure fluctuations, J. Appl. frequencies of a submerged-confined disk in water, J. Sound
Fluid Mech. 12 (5) (2019) 1463–1473. Vib. 337 (2015) 161–180.