12th IFToMM World Congress, Besançon, June 18-21, 2007

A Low Cycle Fatigue Analysis on a Steam Turbine Bladed Disk-Case Study

J.C. Pereira1 L. A. M. Torres2 E. da Rosa3 H. Bindewald4
Mechanical Eng. Tractebel Energia Mechanical Eng. Mechanical Eng.
Dept. – UFSC S.A. Florianópolis, Dept. -UFSC Dept. -UFSC
Florianópolis, Brazil Brazil Florianópolis, Brazil Florianópolis, Brazil

Abstract- This work concerns the root-cause

investigation of the blade damage in a steam
turbine bladed disk of a thermoelectric power
plant. The purpose of this work is to investigate the
low cycle fatigue problem by determining the
natural frequencies and the associated modes of
vibrations of a bladed disk and comparing with the
operational rotation of the turbine and its
harmonics. The bladed disk is an assembly of
several blades grouped and each group is an
assembly of four or five blades, connected by
elements named shroud and tie-wire. The
computational model of a single blade and of a
blades grouped are fitting by modal analysis
measurements. Finally, a numerical modal analysis
is made on a complete bladed disk in order to
investigate the low cycle fatigue problem.
Keywords: steam turbine, blades, fatigue, modal Fig. 1. Overall view of the steam turbine
analysis, finite element modeling

I. Introduction

This paper presents the root-cause

investigation of the blade damage in a steam
turbine bladed disk, operating in a thermoelectric
power plant. This turbine is a part of a turbo-
generator that generates 320 MW in a power plant,
situated in the Santa Catarina state, Brazil, as
shown in figure (1).
Recently, after a forced outage of the plant, a
carefully site inspection in the fifth stage of the
low-pressure turbine has identified that one blade
was broken and several others were in a cracking
Fig. 2. Beach marks on a cross section of a broken blade
process. A preliminary metallurgical examination
of the fractured component showed beach marks,
For the investigation of the blade damage
which is characteristic of a fatigue failure
analysis, one considers both the low cycle fatigue
mechanism, (see figure 2).
failure (LCF) and the high cycle fatigue failure
_____________________________ (HCF) as the mechanism of damage, [1] and [2]. In
1
E-mail: jcarlos@grante.ufsc.br the low fatigue failure analysis, the
2
E-mail: ltorres@tractebelenergia.com.br eigenvalues/eigenvectors of the fifth stage
3
E-mail: darosa@emc.ufsc.br including the centrifugal force effect are
4
E-mail: hbind@pop.com.br determined. In this case, the association between
the eigenvalues and eigenvectors are compared
with the harmonics of the operation speed of the
turbine, and the diametral modes number (nD), see
[3] and [4]. In the high cycle fatigue failure
analysis, the vibratory stress due to the flow-
induced vibrations on the bladed disk is determined
in order to compare with the fatigue limit of the
material, [5], [6] and [7]. In this paper, one
proposes a numerical/experimental procedure to
analyse the low cycle fatigue problem.
The computational mechanics procedure used
in this work consisted first in digitize a single blade
for obtaining a file with the geometric dimensions
and, secondly, to build a solid CAD model of the
blade. The CAD model was verified by comparing
the calculated and measured blade weights. This
model was analyzed using a finite element code in
Fig. 3. Experimental set for the modal analysis of a single
order to obtain the natural frequencies and the blade in a free-free condition
associated vibration modes of a single blade in a
free-free condition. A modal testing was also Figure 4 shows the first mode of vibration and
performed and a numerical/experimental figure 5 shows the second mode of vibration of the
comparison was used to adjust the material single blade in a free-free condition. In the
properties and the geometric dimensions. numerical model the medium size of the elements is
10 mm and the total number of elements is 14 313.
In addition, a five blades grouped was
modeled considering the tie-wire and the shroud as
the elements used to connect the blades. In this
case, another modal testing was performed and the
numerical/experimental comparison was made to
adjust the material properties introduced in the
computational model with the objective of
simulating the contact between the blade and the
tie-wire and the blades and the shroud. As shown in
several papers and reports, these elements can play
a very important role in this type of analysis, [8]. A
combination of modal testing and finite element
(a)
analysis are the current procedure to accurately
predict the dynamic response of bladed disks in
turbo-machinery.

Finally, a fifth stage of the low-pressure

turbine model was built to solve the
eigenvalues/eigenvectors problem and the low
fatigue problem was investigated.

II. Modal analysis on a single blade

The purpose of this analysis is to adjust the

material properties and the geometric dimensions of (b)
the blade in the numerical model, in order to fit the
modal testing measurements. Fig. 4. First mode of the single blade: (a) numerical and
Figure 3 shows the blade configuration for the (b) experimental
modal testing in a free-free condition. The
instruments used for this purpose were a force
transducer B&K 8200, a shock hammer (coupled to
the force transducer), an accelerometer B&K 4344
and the LMS SCADAS III with a Fourier analysis
software. For the mode identification, accelerations
were obtained at 105 points on the blade.
 blade and the shroud is analyzed by using a soft
material between them. So, the purpose of this
analysis is to adjust the material properties,
introduced in these regions in the numerical model,
so that the numerical solution fits the experimental
results.

shroud

(a)

tie-wire blade
blade

root

Fig. 6. Four and five bladed groups

(b)
Fig. 5. Second mode of the single blade: (a) numerical
The experimental set for the modal testing of a
and (b) experimental five blades grouped is presented in figure 7. The
instruments used for this purpose were the same
used in the experimental setup in the modal testing
TABLE I presents the numerical/experimental of the single blade test.
results of the first seven modes of the modal
analysis of a single blade in a free-free condition.

Mode Numerical Experimental Error

(Hz) (Hz) (%)
1 162,8 166,9 -2,5
2 442,3 451,7 -2,8
3 524,6 547,2 -4,1
4 738,7 748,9 -1,4
5 1.112,0 1.077,2 3,2
6 1.212,0 1.166,1 3,9
Fig. 7. Experimental set for the modal analysis of a five
7 1.448,0 1.457,7 -0,7 blades grouped in a clamped-free condition

TABLE I – Numerical/experimental results on a
single blade
III. Modal analysis on a five blades grouped
A tie-wire and a shroud, as seen in figure 6,
connect groups of four and five blades. The contact
problem between the blade and the tie-wire and the
Figure 8 shows the first mode of the 5 blades
grouped in bending and figure 9 presents the
second mode of the 5 blades grouped in bending. In
the numerical model the medium size of the
elements is 7 mm on the blades, 4 mm on the
shroud, 3 mm on the tie-wire and 2 mm on the
material interface. The total number of elements
used in the model is 98 184.
 IV. Modal analysis of the fifth stage

The fifth stage of the low pressure turbine was

modeled in accordance with the results obtained in
the modal analysis on a single blade and on a five
blades grouped. It is composed by 133 blades
divided in groups of four and five blades as shown
in figure 10 (five blades group in gray and four
blades group in black).

Fig. 8. First mode of the five blades grouped in bending

Fig. 10. Fifth stage model of the low pressure turbine

As found by Chyou in [3], one observes two

modal groups with a large modal density. In the
first modal group, blades are associated with first
Fig. 9. Second mode of the five blades grouped in mode of the blades grouped in bending (see figure
bending 8) and in the second modal group, blades are
associated with the second mode of the blades
TABLE II presents the numerical/experimental group in bending (see figure 9).
results of the first eight modes of the modal TABLE III presents the first diametral modes
analysis of a five blades grouped, in a clamped-free in the first group and TABLE IV presents the first
condition. Mode types are bending (B) and/or diametral modes in the second group of the fifth
torsion (T). stage of the turbine turning at its operation speed of
3600 rpm.
Numerical Experimental
The circunferencial modes are not excited by
Freq. (hz) Mode Freq. (hz) Mode Mode type
the force acting on the blades due to steam flow, so
107,8 1 99,4 1 B they are ignored in this analysis.
- - 142,3 2 B/T
172,7 2 148,5 3 B/T
213,6 3 210,7 4 B/T
- - 296,2 5 B/T
395,2 4 336,4 6 B/T
411,0 5 407,6 7 B/T
499,1 6 479,4 8 B/T

TABLE II. Numerical/experimental results on a five

#1 – 1D – 152,70 hz #2 – 1D – 152,8 hz
blades grouped

#3 – 2D – 153,2 hz
#6 – 3D – 154,1 hz
TABLE III. Diametral modes in the first modal group
#31 – 1D – 184,4 hz
#33 – 2D – 185,1 hz
#35 – 3D – 185,9 hz #36 – 3D – 186,2 hz
#4 – 2D – 153,3 hz
TABLE IV. Diametral modes in the second modal group
TABLE V shows the vibrations modes which
can cause the resonance of the fifth stage of the
turbine at 3600 rpm due to the harmonics of the
operation speed and due to the diametral mode
number (nD).
Operation Operation Harmonic/ Frequency
speed speed (hz) Diametral mode
(hz)
(rpm) number (nD)
#7 – 3D – 154,4 hz 3600 60 1 60
3600 60 2 120
3600 60 3 180
TABLE V. Diametral modes in the second modal group
In observation of TABLE III, it is possible to
note that the frequencies in the first mode group are
approximately 150 hz, far from the second and the
third harmonics of the operation speed.
Nevertheless, in TABLE IV, it is observed that the
#32 – 1D – 184,4 hz
frequencies in the second mode group are
approximately 180 hz, near the third harmonic of
the operation speed.
The two 3D diametral modes in the second
mode group 185,9 hz and 186,2 hz have a
deviation of 3,3% and 3,4% from the third
harmonic of the operation speed of the turbine,
respectively. Orsagh and Roemer in [4] recommend
a value of 3% far from one of the harmonics to
assure an operation without resonance. Based on
this results, it is possible to conclude that in the
fifth stage of the low-pressure turbine there will be
#34 – 2D – 185,4 hz no low cycle fatigue problem.
V. Conclusions

In this work, the low cycle fatigue problem was

investigated on a fifth stage of the low-pressure
turbine in a thermoelectric power plant. The
procedure for this study consisted initially in the
fitting of the computational model of a single blade
and a five blades grouped with experimental
measurements. Further, a computational model of
the complete fifth stage bladed disk was developed
and a modal analysis was performed. By analyzing
the harmonics of the operation speed and the
diametral modes of the bladed disk, it was
concluded that no resonance had occurred, so the
low cycle fatigue is not the root-cause of the
damage on the blades. This fact indicates that the
high cycle fatigue on the blades in the fifth stage of
the low-pressure turbine must be investigated.

References

[1] Dewey, R. P. and Rieger, N. F., 1985, Survey of

Blade Failures in Turbine-Generator Units.
Proceedings: Steam Turbine Blade Reliability
Seminar and Workshop, ed. By R. G. Brown and J.
F. Quilliam, EPRI CS-4001, pp. 2-61 ~ 2-78.
[2] Ortolano, R. J., 1991, Recent Case Histories in the
Inspection, Modification, and Repair of Steam
Turbine Blading. Proceedings, International Joint
Power Generation Conference, ASME PWR-Vol.13,
pp. 147-154.
[3] Chyou, Y., 2000, Root-Cause Investigation on Blade
Failures in Steam Turbines From an Aspect of
Computational Mechanics. Proceedings of 2000
International Joint Power Generation Conference,
Miami, USA.
[4] Orsagh, R. F. and Roemer, M. J., 1994, Examination
of Successful Modal Analysis Techniques Used for
Bladed-Disk Assemblies. Technical Report, Impact
Technologies, New York.
[5] Fleeter, S., Zhou, C., Houstis, E. and Rice, J., 1998,
Fatigue Life Prediction of Turbomachine Blading.
Technical Report, Lawrence Livermore National
Laboratory.
[6] Boven, R. and Lam, T. Management Strategy and
Technical Plan Used to Select, Qualify and Procure a
New 20.9” L-1 Stage Steam Turbine Bucket.
Proceedings of 2000 International Joint Power
Generation Conference, Miami Beach, Florida, July
23-26, 2000.
[7] Rao, J. S., Turbine Blade Life Estimation. Alpha
Science International Ltd., 2000.
[8] Shiga, M., Vibration Characteristics of Grouped
Steam Turbine Blades. JSME International Journal,
pp. 592-596, Series III, Vol. 32, No.4, 1989.