The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0264-4401.htm

Dynamic stresses
Analysis of dynamic stresses in Kaplan
in Kaplan turbine blades turbine blades
Lingjiu Zhou
College of Water Conservancy and Civil Engineering, 753
China Agricultural University, Beijing, People’s Republic of China, and
Zhengwei Wang, Ruofu Xiao and Yongyao Luo Received 15 March 2007
Revised 16 July 2007
Department of Thermal Engineering, Tsinghua University, Accepted 20 July 2007
Beijing, People’s Republic of China

Abstract
Purpose – Some comparison of unsteady flow calculation and the measured stress showed that the
dynamic stresses in blades are closely related to hydraulic instability. However, few studies have been
conducted for the hydraulic machinery to calculate dynamic stresses caused by the unsteady hydraulic
load. The present paper aims to analyse the stresses in blades of a Kaplan turbine.
Design/methodology/approach – By employing a partially coupled solution of 3D unsteady flow
through its flow passage, the dynamic interaction problem of the blades was analyzed. The unsteady
Reynolds-averaged Navier-Stokes equations with the SST k-v turbulence model were solved to model
the flow within the entire flow path of the Kaplan turbine. The time-dependent hydraulic forces on the
blades were used as the boundary condition for the dynamics problem for blades.
Findings – The results showed that the dynamic stress in the blade is low under approximately
optimum operating conditions and is high under low-output conditions with a small guide vane
opening, a small blade angle and a high head.
Research limitations/implications – It is assumed that there is no feedback of blade motion on
the flow. Self-excited oscillations are beyond the scope of the present paper.
Originality/value – The authors developed a code to transfer the pressure on blades as a boundary
condition for structure analysis without any interpolation. The study indicates that the prediction of
dynamic stress during the design stage is possible. To ensure the safety of the blades it is
recommended to check the safety coefficient during the design stage for at least two conditions: the
100 percent output with lower head and the 50 percent output with the highest head.
Keywords Turbines, Fluid-mechanics, Stress (materials)
Paper type Research paper

1. Introduction
With the increasing unit capacity and runner diameter, the problem of unsteady flows
in hydraulic turbines has attracted considerable attentions. The energy transfer in a
turbine is accompanied by more or less pressure oscillations caused by vortexes,
cavitations and other complex unsteady flow phenomena in the flow path, which can
cause excessive blade vibration, leading to structural fatigue failures. Most of the Engineering Computations:
International Journal for
researches have focused on pressure fluctuations in the flow path, such as caused by Computer-Aided Engineering and
RSI effects (Guedes et al., 2002; Wang et al., 2002), and vortex rope in the draft tube Software
Vol. 24 No. 8, 2007
pp. 753-762
q Emerald Group Publishing Limited
The authors would like to acknowledge Shuikou Hydropower Plant for presenting the original 0264-4401
data of the research. DOI 10.1108/02644400710833288
EC (Jacob and Prenat, 1996; Bhan et al., 1988; Arpe and Avellan, 2002; Wang and Zhou,
24,8 2006) Most recently a few researchers have measured dynamic stress in the blades
(Pan, 2004). Some comparison of unsteady flow calculation and the measured stress
showed that the dynamic stresses in blades are closely related to hydraulic instability
(Wang and Zhou, 2006). However, few studies have been done for the hydraulic
machinery to calculate dynamic stresses caused by the unsteady hydraulic load. In this
754 present paper, the stresses in blades of a Kaplan turbine were analyzed. The prototype
studied here was a ZZA315-LJ-800 Kaplan (pitch-angle adjustable) turbine with a
specific speed of 22.6, a runner diameter of 8 m, rated output of 204 MW, rated head of
47 m and rated speed of 107 rpm. The stresses in blades were calculated under various
operating conditions to analyze the relationship between the dynamic stresses and
operating conditions.

2. The calculation of stress in blades by coupling solution of fluid and

structure
There are two strategies for the coupled solution of dynamic of fluid and structure
interaction. One is the loosely coupled method with the results of the dynamic forces
and structural displacement exchanged as boundary condition for each other at every
time step (Younsi et al., 2001; Filsinger et al., 2002; Herfjord et al., 1999; Gnesin et al.,
2004). The other is the fully coupled method, the fluid and structure problem are solved
within the same code and the same grids (Slone et al., 2002; Chen and Zha, 2005).
Obviously, only the fully coupled model is rigorous in the physical sense. However, the
calculation based on a fully coupled fluid – structure interaction is CPU-expensive
due to the intensive iterations between the fluid and structure system, especially for
the large-scale calculation for a prototype Kaplan turbine with runner diameter of 8 m.
Here the author adopted the partial coupling concept, which assumes that blade
vibrations causing unsteady effects are significantly smaller than unsteady effects due
to instability of the flow and even neglectable. That is to say, there is no feedback of
blade motion on the flow. This hypothesis is reasonable since self-excited oscillations
are beyond the scope of the present paper.
The flow path was from the spiral casing inlet to the draft tube outlet. The whole
path was discretized with an unstructured hybrid mesh of hexahedron and tetrahedron
cells. Normally the fluid simulation need finer grid than structure analysis. So the mesh
size was determined according to Ma and Zhou’s (2006) recommendation to meet the
requirement for fluid simulation. The final mesh had about 730,000 nodes (Figure 1).

inlet

Figure 1. outlet
Calculation domain for the
fluid and structure
(a) fluid domain of the whole flow path (b) solid domain of the blade structure
In the course of the calculations, the runner zone mesh rotated at 107 rpm with the Dynamic stresses
relative positions of each node in this zone unchanged. The interfaces between the
guide vanes, the runner and the draft tube were modeled with sliding interfaces with
in Kaplan
the flux transferred between neighboring zones. turbine blades
The water was considered to be incompressible. A time-dependent Reynolds
average Navier-Stokes (RANS) simulation was performed to calculate the flow field
in the Kaplan turbine path. The continuity equation and the momentum equations 755
for the fluid within static frame (spiral case, stay vanes, wicket gates and the draft
tube) are:
›
ðu~ i Þ ¼ 0 ð1Þ
›xi


›u i › ›p › ðm þ mt Þ ›ui ›uj
þ ðui uj Þ ¼ 2 þ þ þ fi ð2Þ
›t ›x j r›xi ›xj r ›xj ›xi
where p is the pressure, r the density, ui the component of velocity vector u~ ðu1 ; u2 ; u3 Þ
and fi the component body force vector fð ~ f 1 ; f 2 ; f 3 Þ: m the laminar viscosity, mi the
turbulent viscosity which was closed by SST k-v turbulence model (Menter, 1994).
The continuity equation and the momentum equations for the fluid in the rotational
frame (runner) are:
›
ðwi Þ ¼ 0 ð3Þ
›xi


›w i › ›p › ðm þ mt Þ ›wi ›wj
þ ðwi wj Þ ¼ 2 þ þ þ f 0i ð4Þ
›t ›xj r›xi ›xj r ›xj ›xi
~ 1 ; w2 ; w3 Þ: f 0i the component of
where wi is the component of relative velocity vector wðw
~
force vector F .0

The two velocity vectors u~ ; w

~ are related by:
u~ ¼ w
~ þ v~ £ ~r ð5Þ
Force vector F~ 0 is written as:

~ 2 v~ £ ðv~ £ ~rÞ þ f~
F~ 0 ¼ 22v~ £ w ð6Þ
Here v~ is the angular velocity of the runner and ~r is the position vector in the rotation
frame. Only gravity was considered in vector f~.
The time-dependent RANS model was discretized using the control-volume
technique through the SIMPLEC scheme with a second-order upwind scheme used for
the convection terms and a central difference scheme for the diffusion terms in the
momentum equations. The time step was 0.0056 s, which is 1/100 of the runner
rotational period. This time step was validated to be enough for catching some main
pressure fluctuation frequencies such as rotational frequency and blade passing
frequency.
The pressure conditions were set on inlet and outlet according to the head of
each operation condition and estimated velocity on the two boundaries. The initial
EC turbulence variables on the inlet and outlet were estimated according to the hydraulic
diameters and turbulence intensity, which was set to be 6 percent on the inlet and
24,8 7 percent on the outlet.
The structure mesh is shown in Figure 1. The mesh near blade pivot was carefully
generated to ensure the accurate representation of the fillet corner in order to avoid
false stress concentration. The transient dynamic equilibrium equation for the stresses
756 in a linear structure is:
½M
{u}
€ þ ½C
{u}
€ þ ½K
{u} ¼ {F t } ð7Þ
where [M ] is the mass matrix, [C ] the damping matrix and [K ] the stiffness matrix,
{u} i the nodal displacement vector, {u} € the nodal velocity vector and {u€ } the nodal
acceleration vector, {F t } the node load vector, including gravity, centrifugal force and
pressure force that was transferred from flow analysis.
Equation (7) was discretized using the FEM method and solved using the Newmark
method. The iterative equation was (Hallquist, 1998):


1 d
½M
þ ½C
þ ½K
{u nþ1 } ¼ {F t } þ ½M

aDt 2 aDt


 
1 1 1
 {u n } þ {u n } þ 2 1 {u€n þ ½C

} ð8Þ
aDt 2 aDt _ 2a



 
d d Dt d
 {u n } þ 2 1 {u_n } þ 2 2 {u€n }
aDt a 2 a
where a and d are the Newmark integration parameters and Dt ¼ tnþ1 2 tn :
Equation (8) shows that the effects of the velocity and acceleration terms are
considered in the transient displacement computation. {u} was obtained from
equation (8) with {s} calculated at every time step from:
fsg ¼ ½D
½B
{u} ð9Þ
where [D] is the elastic matrix based on Young’s modulus and Poisson’s ratio for the
material, [B] the strain-displacement matrix based on the element shape functions.
The dynamic stresses in blades were computed based on the fourth strength theory.
The von Mises or equivalent stress, se, was computed using:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 iffi
1h 2 2 2

2 2 2
se ¼ ðsx 2 sy Þ þ ðsy 2 sz Þ þ ðsz 2 sx Þ þ 6 txy þ tyz þ txz ð10Þ
2
The blade material is cast steel Cr13Ni5Mo with the properties listed in Table I.
The key problem was to transfer the pressure load p on the blades to the dynamic
analysis code. To do this the grids on blade surfaces and the hub were identical for both
fluid mesh and structure mesh. A small code was developed to generate an index for the
interface nodes in the two domains for the purpose of transferring data at each time.

Density (kg/m3) Young’s modulus (Pa) Poisson’s ratio

Table I.
Blade material properties 7.85 £ 103 2.06 £ 1011 0.3
The advantage of this method is that the pressure on the blades can be precisely Dynamic stresses
transferred without any interpolation. The flow field simulation started first. For most in Kaplan
cases, the frequencies and amplitudes of the pressure oscillations on the monitoring
nodes changed very little after several rotational periods. Then the structure calculation turbine blades
was conducted and the calculation results were recorded for several periods.

757
3. Calculation result
The calculations were performed for 13 operating conditions with operating
parameters listed in Table II.
During the calculations, two points on one blade pressure side ( p1) and the suction
side (s1) were selected as monitoring points (Figure 2). Pressure fluctuations at the two
points are shown in Figure 3 for three typical conditions GK3, GK8 and GK10.
Under the conditions approximating the optimum point and those with large blade
angles and guide vane openings, the pressure on the blade fluctuates with the runner

Net head H Power N Guide vane opening GVO Blade angle BA

Operating condition (m) (MW) (percent) (8)

GK1 61 235 65.6 17.29

GK2 61 215 62.4 15.44
GK3 61 100 36.21 5.57
GK4 57.3 215 65.76 16.68
GK5 57.3 100 40.54 6.16
GK6 53 235 78.58 20.41
GK7 53 215 74.05 18.3
GK8 53 150 59.03 11.72
GK9 53 100 47.19 6.95
GK10 43.8 235 94.6 25.82
GK11 43.8 215 90.46 23.32
GK12 43.8 100 57.37 9.43 Table II.
GK13 38 180 97.66 23.26 Calculated conditions

Figure 2.
Recording points on one
P1 S1 blade with (a) viewed from
runner inlet to outlet;
(b) viewed from runner
outlet to inlet
(a) Recording point p1 on pressure side (b) Recording point s1 on suction side
EC 420 pressure on pl, GK3 356.5 pressure on pl, GK8
pressure on pl, GK10
188
24,8 400
356.0
P/kPa

P/kPa

P/kPa
184
380
355.5
360 180
355.0
758 16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5
t/s t/s t/s
(a) (b) (c)
pressure on sl, GK3 pressure on sl, GK10
15 pressure on sl, GK8 –145
–50.5
10
–51.0 –150
5
P/ kPa

P/ kPa

P/ kPa
–51.5
0
–155
–5 –52.0

–10 –52.5 –160

16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5
t/s t/s t/s
(d) (e) (f)
turque on one blade, GK10
960 turque on one blade, GK3 760 turque on one blade, GK8
800
920 755

T/ kN.m
T/kN.m

T/kN.m

880 760
750
Figure 3. 840
745
Pressure fluctuations and 800
720
torque on one blade for 740
16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5
three typical operating
t/s t/s t/s
conditions
(g) (h) (i)

rotation frequency, fn, as shown in Figures 3(b), (e), (c) and (f) for conditions GK8 and
GK10. However, for conditions with small blade angles and small guide vane openings,
the blade pressure fluctuations are accompanied by increased turbulence and the
fluctuations which does not show any obvious periodic feature, as shown in Figure 3a
and d for condition GK3. Figure 4 shows that the randomness pressure fluctuations in
GK3 are related to the secondary flow across the flow path. There are obvious vortexes
near hub within no blade region between the guide vanes and runner blades for GK3
with small guide vane opening. The random movements of vortexes cause pressure
fluctuations on blades. For the same reason, the pressure fluctuation amplitude for
GK3 is greater than for the other two conditions.
Figure 3 also shows that the mean value and amplitude of the torque fluctuations is
small for conditions near the optimum point (GK8, Figure 3h), is larger for large blade
angles and large guide vane openings (GK10, Figure 3i) and even larger for small
blade angles and small guide vane openings (GK3, Figure 3g), which was verified by
the power station’s operating recordings. When the turbine was operating under
approximately optimum condition (for example, GK8) the operating oil pressure inside
the piston chamber was normal, but operating under approximate condition GK3 the
oil pressure inside the chamber was extremely high.
 Dynamic stresses
in Kaplan
turbine blades

759

(c) GK10
(b) GK8
(a) GK3

Figure 4.
Velocity vectors in no
blade region between the
guide vanes and runner
blades
EC Typical pressure distribution and the von Mises static stress distributions in blades are
24,8 shown in Figure 5 for condition GK3. The maximum stress is near the blade root for all
cases. This was consistent with general knowledge on static stress distributions in
Kaplan turbine blades. The dynamic stresses at the node with maximum stress are
plotted in Figure 6 for the three typical conditions.
Figures 3 and 6 show that the dynamic stresses in blades are closely related to the
760 pressure fluctuations in the flow path. For most cases, the main pressure fluctuation
frequency and dynamic stress frequency is the runner rotation frequency fn. The mean
stresses and the dynamic stresses amplitudes (with 95 percent probability) at the node
with the maximum stress are shown in Figure 7 for all calculated conditions. The mean
stress increased with head and decreased with blade angle. This is due to the fact that

Pressure
5.792e+005

3.959e+005

2.126e+005

2.931e+004
Figure 5.
Pressure distribution on
–1.540e+005
blades and stress
[Pa] 154895 0.294E+08 0.586E+08 0.879E+08 0.117E+09
distributions in blades on 0.148E+08 0.440E+08 0.732E+08 0.102E+09 0.132E+09
condition GK3
(a) Pressure distributions (Pa) (b) Stress distributions (MPa)

136 GK8 104 GK10

GK3
120.2
se /MN.m

132
se /Mpa

se /kPa

102
120.0
Figure 6. 128
Dynamic stress at the 119.8
100
node with the maximum 124
16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5 16.5 17.0 17.5 18.0 18.5
stress (condition GK3,
t/s t/s t/s
GK8, GK10)
(a) (b) (c)

Head = 61 m Head = 61 m
140 Head = 57.3 m 140 Head = 57.3 m
Head = 53 m Head = 53 m
130 Head = 43.8 m 130 Head = 43.8 m
Head = 38 m Head = 38 m
se / MPa

se / MPa

120 120
Figure 7. 110 110
The maximum mean
stress levels and dynamic 100 100
stresses amplitudes
(plotted with error bars) 4 8 12 16 20 24 28 40 60 80 100
for 13 conditions
Blade angle (°) Guide vane opening (%)
the overall pressure difference between the pressure side and suction side of one blade Dynamic stresses
increases with head, and the bending effects increase with the decrease of blade angle. in Kaplan
Dynamic stress amplitude is low within 60-80 percent opening, and is higher at high
opening of above 90 percent with low head (GK10, GK11, GK13) and is the highest at turbine blades
low opening (36 percent) with high head (GK3).
The maximum dynamic stresses amplitude is 8.6 MPa, which is less than the
limited stress of blade material. So blades are safe in practical application. In fact, the 761
runner has been operating for more then ten years without any cracks.

4. Conclusions
The dynamic stresses in the blades of a Kaplan turbine were analyzed for 13 operating
conditions using the CFD analyses in flow system coupled with the stresses analysis
in the runner. The stress distributions in blades show that the highest stress was
located near the blade root. The mean stress increases with head and decreases with
blade angle. The dynamic stresses in blade are closely related to the pressure
fluctuations inside the flow path, which are low within 60-80 percent opening, are
higher at 90 percent opening (full output) with low heads and are the highest at low
opening (36 percent) with high head. The present study indicates that the prediction of
dynamic stress during design stage is possible. To ensure the safety of blades it is
recommended to check up safety coefficient during design stage at least for two
conditions, the 100 percent output with lower head and the 50 percent output with the
highest head.

References
Arpe, J. and Avellan, F. (2002), “Pressure wall measurements in the whole draft tube: steady and
unsteady analysis”, Proceedings of the 21st IAHR Symposium on Hydraulic Machinery and
Systems, Lausanne, Switzerland.
Bhan, S., Codrington, J.B. and Mieke, H. (1988), “Reduction of Francis turbine draft tube surges”,
paper presented at Fifth International Symposium on Hydro Power Fluid Machinery,
New York, NY, pp. 95-101.
Chen, X.Y. and Zha, G.C. (2005), “Fully coupled fluid-structural interactions using an efficient
high resolution upwind scheme”, Journal of Fluids and Structures, Vol. 20, pp. 1105-25.
Filsinger, D., Szwedowicz, J. and Schafer, O. (2002), “Approach to unidirectional coupled
CFD-FEM analysis of axial turbocharger turbine blades”, Transactions of ASME: Journal
of Turbomachinery, Vol. 124 No. 1, pp. 125-31.
Gnesin, V.I., Kolodyazhnaya, L.V. and Rzadkowski, R. (2004), “A numerical modelling of
stator–rotor interaction in a turbine stage with oscillating blades”, Journal of Fluids and
Structures, Vol. 19, pp. 1141-53.
Guedes, A., Kueny, J.L., Ciocan, G.D. and Avellan, F. (2002), “Unsteady rotor-stator analysis of a
hydraulic pump-turbine – CFD and experimental approach”, Proceedings of the 21st IAHR
Symposium on Hydraulic Machinery and Systems, Lausanne, Switzerland.
Hallquist, J.O. (1998), LS-DYNA Theoretical Manual, Livermore Software Technology
Corporation, Livermore, CA.
Herfjord, K., Drange, S.O. and Kvamsdal, T. (1999), “Assessment of vortex-induced vibrations on
deepwater risers by considering fluid-structure interaction”, Journal of Offshore
Mechanics and Arctic Engineering, Transactions of the ASME, Vol. 121 No. 4, pp. 207-12.
EC Jacob, T. and Prenat, E. (1996), “Francis turbine surge: discussion and data base”, paper
presented at XVIII IAHR Symposium, Dordrecht, pp. 855-64.
24,8 Ma, W.S. and Zhou, L.J. (2006), “The effect of grid on the result of CFD simulation of turbine”,
Journal of Hydroelectric Engineering, No. 1, (in Chinese), pp. 72-5.
Menter, F.R. (1994), “Two-equation eddy-viscosity turbulence models for engineering
applications”, AIAA Journal, Vol. 32, pp. 1598-605.
762 Pan, L.P. (2004), “Study on runner stress measurement of large-capacity turbine”, dissertation,
Tsinghua University, Beijing (in Chinese).
Slone, A.K., Pericleous, K., Bailey, C. and Cross, M. (2002), “Dynamic ?uid-structure interaction
using finite volume unstructured mesh procedures”, Computers &Structures, Vol. 80,
pp. 371-90.
Wang, Z.W. and Zhou, L.J. (2006), “Simulations and measurements of pressure oscillations
caused by vortex ropes”, Journal of Fluids Engineering, Vol. 128 No. 4, pp. 649-55.
Wang, Z.W., Zhou, L.J. and Huang, Y.F. (2002), “The rotor-stator interaction flow simulation on
three gorges hydraulic turbines”, Proceedings of the 21st IAHR Symposium on Hydraulic
Machinery and Systems, Lausanne, Switzerland.
Younsi, R., Batanony, I., Tritsch, J.B., Naji, H. and Landjerit, B. (2001), “Dynamic study of a wind
turbine blade with horizontal axis”, European Journal of Mechanics – A/Solid, Vol. 20,
pp. 241-52.

Corresponding author
Zhengwei Wang can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints