Proceedings of ASME Turbo Expo 2009: Power for Land, Sea and Air

GT2009
June 8-12, 2009, Orlando, Florida, USA

GT2009-59040

STEAM TURBINE LOW PRESSURE BLADES FATIGUE FAILURES CAUSED BY

OPERATIONAL AND EXTERNAL CONDITIONS

Zdzislaw Mazur* Rafael García-Illescas Alejandro Hernández-Rossette

Instituto de Investigaciones Instituto de Investigaciones Instituto de Investigaciones
Electricas Electricas Electricas
Cuernavaca, Morelos, Mexico Cuernavaca, Morelos, Mexico Cuernavaca, Morelos, Mexico

ABSTRACT of blades in turbines. Flow-induced blade oscillations (flutter)

This paper provides an overview of a steam turbine low of the turbine can lead to fatigue failures of a construction and
pressure blades failures induced by flow excitation and blades so they represent an important problem of reliability, safety, and
torsional vibrations due to sudden changes on the grid. The operating cost. Aeroelasticity phenomena can occur while
analysis include L-0 and L-1 blades failures of the 110 MW [1], turbine is operating at low load/low vacuum conditions
28 MW [2] and 35 MW [3] geothermal units induced by (LL/LV). During turbine operation at a LL/LV, the L-0 and last
unstable flow due to operation at low load low vacuum, and L-0 but one (L-1) blade vibration stresses are increasing abruptly.
blades failure of the 660 MW [4] nuclear unit due combined This increase (peak) of the blade vibration is induced by
effect of a transient phenomenon provoked by sudden load unstable flow with oscillation of a shock wave near the throat of
changes on the grid and some low vacuum operation period. the blade tip passage [5, 6]. Prediction of the forced response of
The failure initiation was registered at different zones of the rotor disc assemblies is still a challenging engineering task
blades depending on the case, and were localized at the blades because of unknown excitation loads and friction damping
cover segments, blade root and blade airfoil close to the effects [7]. Recently, a remarkable progress in transient flow
platform. Laboratory evaluation of the blades fracture surface calculations allowed the prediction of more realistic excitation
indicates the failure mechanism to be high cycle fatigue (HCF). forces acting on the rotating blades [8, 9, 10, 11, 12, 13].
Nevertheless, due to some current uncertainties in transient flow
KEYWORDS calculation and forced response of rotor disc assemblies, as was
Failure analysis; High cycle fatigue; Low-pressure blades reported herein, some cases of steam turbine free standing and
failure; Metallurgical examination; Steam turbine failures; continuously-coupled low pressure blades failures have been
Torsional vibration, Low load operation. recorded. According to [2 and 14], the steam turbine operation
with LL/LV induces last-stage (L-0, L-1) blade excitation
1 INTRODUCTION (vibration) by unstable flow developing high vibratory stresses.
The last stage blade (L-0 blade) is one of the most Due to reduced mass flow, the steam conditions are variable
important contributors to the performance and reliability of the along the steam path; there are zones of different pressure,
steam turbine. With the last stage blades typically producing radial flows, counter flows, flow recirculation (flow
10% of the total unit output, and up to 15% in some combined- instabilities). These operation conditions generate blade
cycle applications, improvements in last stage efficiency can excitation forces, which can lead to blade failures. Other
significantly impact the output of the total unit. Retrofitting an problem which exists in turbine and generator (t/g) rotor
older last stage design with a modern diaphragm and last stage components are torsionally induced high cycle vibration. Two
blade can typically improve heat rate and output by up to 1% types of changes in the turbo generator (t/g) system can cause
[5]. Modern turbomachines operate under very complex torsional variation: (a) turbine perturbations and (b) electrical
regimes where a mixture of subsonic, transonic and supersonic grid perturbations. The electrical perturbations are due to
regions coexists. The trend for improved large steam turbine sudden changes on the grid such as large motor starts and arc
design towards higher aerodynamic blade loading and smaller furnace operation. In general, there are always relatively small
physical size attracts much attention to the aeroelastic behavior perturbations that create a sort of broadband background noise,

*Investigator and author of correspondence 1 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 but with large equipment, such as arc furnaces, the perturbations damage of cover segment No. 19; heavy deformation, rubbing,
may be in the order of 20 to 100 MWs with similar VAr fracture and separation of the pieces of material are apparent. In
changes. These transients may cause phase imbalance and Figure 2b the crack initiation on cover segment No. 12 is
negative sequence currents [15]. Torsional vibration of rotors, shown. The crack is localized on the outer filet radius of the
source of fatigue, is generally a sporadic, transient phenomenon wall of the cover segment hole. There were more cover
provoked by sudden load changes on the grid and/or inter- segments with similar cracks.
harmonic loading which lasts from seconds to minutes [15].
Most of the time these transient events do not overly excite the
t/g shaft torsional resonances, but in the case of a coincidence
of the transient’s wave form characteristics (for example
operation periods with LL/LV) and torsional resonances, it is
resulting in several cycles of high stress. The accumulation of
these cycles may lead to crack initiation and fatigue failure.
This paper provides an overview of some cases of last-stage and
last but one stage blades failure investigation due to flow
excitation (flutter) caused by operation periods with LL/LV and
other case of identification of the combined effect of torsional
vibrations near 120 Hz and operation periods with low load/low
vacuum as the primary contributors to the L-0 blades failure. Figure 1. The group of 12 blades, L-0 row from the generator
side bent at the tip.
2 BLADES FAILURE ANALYSIS

2.1 Failure analysis of last stage (L-0) turbine blades

of a 110 MW geothermal unit.
Background
The blade under evaluation was the 23-inch/3600 rpm last
stage blade (L-0) of a 110 MW geothermal turbine which
consists of two tandem-compound intermediate/low-pressure
turbines with steam condition 1.1 MPag/182 °C/84 kg/s. During
the unit last overhaul (January 2004), the original grouped rigid
L-0 blades were replaced by flexible blades continuously
coupled 360 degrees around the row by loose cover segment at
the tip and loose sleeve and lug at the mid-span. The evaluation
of these failed new installed blades was carried out after one
year of blade operation period in base load mode. The blade is
made of AISI 410 stainless steel. During unit normal operation
period the water feeding pump was tripped due to a mechanical
problem and as a result the unit vacuum was dropped from 648
mmHg to 570 mmHg and the load from nominal to 72 MW (65
% of load) at an interval of 42 seconds from the last stable
point. After the next 26 seconds the unit was tripped due to a Figure 2. Crack initiation on the cover segment No. 12.
condenser low vacuum condition. After the turbine was
restarted there were measured high vibrations, which forced the Metallurgical investigation of the L-0 blade
unit to be shut down to carry out an inspection and related The metallurgical investigation of the failed L-0 blade
maintenance. The turbine visual examination revealed that the included metallography, SEM (scanning electronic microscopy)
group of 12 L-0 blades from the generator side of rotor fractography and chemical analysis. The microstructure of the
connected to the generator, was bent in the direction opposite to blade airfoil (bended tip zone) consists of tempered
the rotor rotation, as shown in Figure 1, and that another group homogenous martensite, free of cracks, typical for forged
of 4 blades of the same row at 140 degrees from the first stainless steel according to specification AISI 410.
damaged group was also bent. The blade cover segments were Fractography evaluation was carried out on the exposed crack
separated from the damaged blades and they present damage in surface of cover segment No. 12 (see Figure 3) using SEM to
the form of bending, rubbing, loss of material and cracks. Also, determine the origin of the fracture. Figure 3 shows the different
the loose lashing sleeves which couple the blades in the mid- zones of the fracture propagation surface of cover segment No.
span were separated from the damaged blades, crushed and 12. The presence of striations (fracture sliding planes) which
fractured in separate parts. Figure 2a shows the detail of the

2 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 are characteristic for fatigue mechanism of fracture propagation, determinated. The maximum stresses, of 391 MPa, were
are noticeable. registered at the airfoil mid-span below the lashing sleeves. The
stress level at the hole of the cover segment in the contact zone
between the blade tenon and the cover segment (fracture zone)
FractureFracture
initiation
initiation
zone zone
was lower; 250 MPa, as is indicated by arrow in Figure 5.
Fracture propagation
Fracture propagation
zone-striations
zone-striations First mode

200 X 1000 X

Figure 3. Microstructure corresponding to cover segment

Second mode
number 12 made of Titanium Ti-6Al-4V.
Each striation represents one cycle of fatigue. Average inter-
striation distance was 3.73 µm in the fracture propagation zone
and 1.70 µm to 1.97 µm in the fracture initiation zone. On the
fracture surface no beach marks were found. The presence of Third mode
beach marks on the fracture surface commonly indicates that
more events participated in fatigue propagation. The beach
marks divide the fracture surface in zones of different
roughness, which correspond to different events of fatigue. The
absence of beach marks on the fracture surface of cover Fourth mode
segment means that only one event participated in fatigue
propagation. The chemical analysis of the deposits present on
the fracture surface of cover segment No. 19 revealed some Figure 4. The modeshapes of continuously coupled blades.
quantity of Ni, Cu, Co and S. The existence of significant
clearances between the cover segment hole and the blade tenon
facilitate accumulation of the deposits (oxides) which come
from other sections of the turbine and can promote corrosion.

Blade stress analysis

Using the Finite Element Method (FEM), centrifugal
stresses, deformations and modeshapes of the individual blade
and coupled blades at 3600 rpm were calculated. The
calculation also included the determination of stresses at the
cover segment. The single blade numerical model is composed
by 2150 nodes and 287 solid elements of quadratic interpolation
with one intermediate node at each edge. Cover segment
numerical model is composed by 1146 nodes and 221 solid
elements of the same characteristics. The static and dynamic
analyses were accomplished using models of 5 and 6 grouped Figure 5. Stress distribution at the cover segment (static
blades. Additionally, dynamic analysis also included cyclic analysis).
symmetry characteristics in order to take to account the Fracture propagation analysis of the cover segment
structural stiffness and mass effects of the disk and blades A fracture propagation analysis of the cover segment was
dynamic interactions. The results were evaluated considering carried out, based on the metallographic investigation findings
the maximum stresses developed at the blade and cover and rules of fracture mechanics. Stable fatigue fracture
segment, maximum deformation and possible resonances. The propagation is governed by Paris Law [16] according to
natural frequency of coupled blades is represented in Figure 4. equation (1).
The first natural frequency is 115 Hz, the second is 195 Hz, the
da
third is 357 Hz and the fourth is 358 Hz. It can be seen that the  C (K ) m (mm/cycle) (1)
third and fourth blade natural frequencies are very similar. The dN
reported natural frequencies of coupled blades confirm that no where:
mechanical resonances of the blades structure exist. The da
centrifugal stress distribution at the blade airfoil was - Velocity of fracture propagation
dN

3 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 da – Crack size increment during one cycle of fatigue da
dN - Number of fatigue cycles a   f  t [mm] (4)
C and m - Empiric constants dN
K – Range of stress intensity factor during fatigue cycles It was found that for frequency f = 333 Hz and striations
distance da = 3 m the fracture length was a = 7.5 mm, which
The fatigue striations distances on the cover segment fracture is very close to the real cover hole depth (a = 7 mm)
surface correspond to the crack size increment, da, during one considered in this analysis. This frequency (333 Hz) is very
cycle of fatigue. These distances were measured and its values close to the natural frequencies of the third and fourth modes of
fall within the range 1.7 m at the fracture initiation zone to 4 vibration of the coupled blades (357 Hz and 358 Hz
m at the fatigue fracture end zone. It was observed that the respectively). Because the striations distance 3 m was found
predominant inter-striation distance was between 3 m and 4 predominantly on the cover segment fracture surface, it can be
concluded that probably the third torsional mode X (Fig.4),
m. To successfully separate the cover segment completely
which is typically related to the turbine operation with low
from the blade, the maximum crack size should be more or less
LL/LV, may be responsible for the cover segment fatigue failure
the same as a depth of the cover segment hole in which the
and its separation from the blades.
blade tenon is installed (a=7 mm). The total time of the
fatigue event was 26 seconds, which correspond to unit low
Discussion of the results
load, low vacuum operation (72 MW, 570 mm Hg). This time
The results of the metallographic examination of the failed
includes the periods of fracture initiation and propagation, the
L-0 blade and cover segment indicate that the crack initiation
time required for the deformation and separation of the cover
and propagation on the cover segment was driven by a high
segment, and also the time required for the blade airfoils to
cycle fatigue mechanism. Striations characteristic to high cycle
deform. Considering arbitrarily that the time for fracture
fatigue were found throughout the whole fracture surface of the
propagation lasted 7 seconds, the possible blade excitation
cover segment. Also, whole fracture surface presented small
frequency that could lead to the cover segment fatigue failure
transgranular cracks, typically related to fatigue fracture
was calculated from equation (2).
mechanisms. Beach marks were not encountered on the cover
a t segment fracture surface. The lack of beach marks on the
f  [Hz] (2)
da dN fracture surface means that only one fatigue event participated
in fracture propagation. Considering the L-0 blades real
where:
operation period (one year), it may be concluded that the
f – Blade excitation frequency
mechanical resonance of the blades does not contribute to blade
a = 7 mm; depth of the cover segment hole
failure. This conclusion also indicates that fatigue failure of the
t = 7 s; time of the fracture propagation
blades/cover segments was not originated during continuous
da/dN = 1.7, 3 and 4 m/cycle; velocity of fracture propagation. operation under vibration stresses, but during transition events.
Using a rearranged form of the Paris Low the stress intensity If the fatigue initiation and propagation were during continuous
factor, K, was determined from equation (3): operation under resonance vibratory stresses, the blade failure
da 1 would occur practically immediately (after a few hours of
K  m [MPa√m] (3) operation). Analyzing the unit’s operation history since the date
dN C
in which the L-0 blades were installed (retrofit), only one period
where:
of unit operation with LL/LV lasting 26 seconds approximately,
C = 2.66e-12 and m = 4.23 determined using nCode [16].
was detected; this is congruent with metallographic findings on
the cover segment fracture surface (lack of beach marks). The
Table 1 shows calculated stress intensity factors and associated
steam turbine operation with LL/LV is inducing L-0 blade
excitation frequencies for three options of striations distances
excitation (vibration) by unstable flow developing high
(da = 1.7, 3 and 4 m). Stress intensity factors can also be
vibratory stresses. Figure 6 shows steam flow stream lines
obtained from the finite element model and a stress calculation.
distribution at the L-0 stage during LL/LV operation obtained
In this case, these data are only for reference.
from Computational Fluid Dynamics (CFD) analysis. Due to
reduced mass flow, the steam conditions are variable along the
Table 1. Stress intensity factors and excitation frequencies.
steam path; there are zones of different pressure, radial flows,
Parameter da [m] counter flows, flow recirculation (flow instabilities). These
1.7 3 4 operation conditions generate blade excitation forces (torsional
K [MPa√m] 23.57 26.96 28.86 vibrations X mode), which can lead to blade failures. Unit
f [Hz] 588.23 333.3 250 operation with reduced mass flow is also causing a reduction of
Further, considering these frequencies and the same options of the flow velocity in the same degree. This results in changes of
striation distance, the length of fracture was calculated using the blade entry flow incidence angle (change of stage velocity
equation (4). triangle); the flow is entering into the L-0 blades with negative

4 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 incidence angle (in this case it was 33-36 ° approximately) stress analysis included stresses developed by centrifugal force,
striking the suction surface of the blade airfoil and exciting the steam flow pressure and combination of both.
blades.

Fracture
initiation 45 mm

Erosion

Figure 6. Steam flow stream lines distribution at the L-0

stage during low load operation.
The pressure fluctuation, flow recirculation and counterflows, Applying blade rotational velocity ω = 376.99 rad/s (3600 rpm)
in conjunction with the negative incidence angle flow striking it was calculated centrifugal acceleration and consider blade
on the blades, developed excessive vibratory stresses causing distributed mass, displacements, unitary deformations and
fatigue fracture and seperation of cover segments. In turn, it stresses were calculated. The blade stresses developed due to
caused blade structural loosening and drastically changed the steam flow pressure were calculated using the results of
blade vibration damping characteristics; the blades operate as it thermodynamic calculations of the radial pressure distribution
they were free-standing. Furthermore, the forces developed by on the blade at 21 flow stream lines. The Von Misses combined
the steam flow and the vibratory stresses finally caused the stresses contours due to centrifugal and steam pressure forces
deformation of the blades. are shown in Figure 8. The maximum stresses reached a value
of 569.2 MPa and are localized in the blade trailing edge
2.2 Failure analysis of last stage (L-0) turbine blade (failure zone) and leading edge as shown in Figure 9. The
of a 28 MW geothermal unit. maximum principal stress was also calculated which is around
Background +574 MPa. The blade maximum stress is lower than yield
The blade under evaluation was the 25-inch/3600 rpm last stage strength of the blade material (762 MPa) but the safety factor is
blade (L-0) of a 28 MW geothermal turbine. The blade was moderate (1.33 to 1.34). It means that there is a small margin to
made of AISI 420 stainless steel. This unit has one flow accommodate possible steam flow excitations of the blade,
intermediate/low-pressure turbine composed by nine stages. The which commonly occur during transitions and flow instabilities
L-0 row consists of 62 free-standing blades. The turbine (LL/LV operation or no load operation).
accumulated of 59,700 operation hours to failure. Figure 7
shows the detail of the fracture of the blade #38 which has
longitude about 45 mm (maximum length of fracture
registered). The fracture initiated at the trailing edge in the zone
of transition radius airfoil-root platform. In this zone it can be
seen blade erosion (it is noted dark spot). In total from 62
blades of the stage 9 in 37 blades (60 %) were registered similar
fractures with varied fracture lengths; from 3 mm to 45 mm.

Blade stress analysis

The blade was measured using 3D optical system to obtain
its topology which was exported to Finite Element Method
(FEM) software and blade numerical model was elaborated.
Blades fir-tree roots are not modeled. Blade numerical model is
composed by 13479 nodes and 2576 solid elements of quadratic
interpolation with one intermediate node at each edge to be able
better modeling of the blades curved zones. The blade static
Figure 8. L-0 blade combined stresses due to centrifugal and
steam pressure forces (MPa).

5 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Mode 2

Mode 1 Mode 3
a) b)

Figure 9. Detail of L-0 blade combined Von Misses stresses.

For value of stressesa)see Fig. 8. a) - Stresses at leading
b) edge, b)-
stresses at trailing edge (failure zone).

Material properties were considered as functions of temperature

Mode 2 Mode 4
which was around 170°C at this last stage. This temperature
value is so low that it doesn’t provoke material degradation. Figure 10. Normalized vibratory stresses in the blade trailing
adge (Faliure zone).
Dynamic analysis of the blade
To analyze blade dynamic performance the modal The vibratory stresses of the first four blade vibration modes in
frequency of the grouped blades for ten natural frequencies these conditions for the blade failure zone (trailing edge) are
were determined. It was found that the fourth mode of blade shown in Figure 11. This figure also shows the vibratory
Mode 3 VIBRATORIOS EN LA ZONA DE FALLA
vibration, which has a frequency of 537.6 Hz is very close to stresses of the blade induced
ESFUERZOS
by resonance of ninth harmonic of
(C/Cc=0.02)
ninth harmonic of the rotor speed (540 Hz). Consider that exist rotor speed indicated by arrow (540 Hz, 2.89 MPa).
some uncertainty in calculation of blade natural frequencies it 70

may mean that blade fourth mode of vibration is in resonance 60

with ninth harmonic of rotor speed and could be related to the
VON MISES (MPa)

50
fracture.
40

30
L-0 blade vibratory stresses analysis Mode 4
On the base of modal analysis which provided L-0 blade 20

natural frequencies, blade dynamic stresses distribution for each 10

modal shape were obtained. The first three modeshapes are 0
mainly bending modes while the fourth one is a torsional mode. 0 100 200 300 400 500 600

In Figure 10 the normalized vibratory blade stresses are FREQ. (Hz)

shown, zones of high vibratory stresses qualitatively are Figure 11. Vibratory stresses in the zone of blade failure (blade
indicated. It can be seen from Figure 10 that for the first four trailing edge) of first four blade vibration modes.
vibration modes the vibratory stresses are concentrated in the
blade trailing edge (failure zone indicated by arrow). The Fatigue analysis of the blade was carried out applying criterion
biggest stresses are developed in the first and second modes of Stress-Life (S-N) [16, 17] which is adequate for the case
vibration and they diminish substantially for the subsequent analyzed due to the very low level of stresses, and consider
modes of vibration. The maximum vibratory stresses occur damping factor C/Cc = 0.02 [18]. The variable stress amplitude
principally in the first mode which is flexural, but there are also of 2.89 MPa at 540 Hz (Figure 11) as the alternating stress and
dynamic stress concentrations in modes 2 a 4. It means that in the centrifugal and steam pressure as a mean stress, were used
the case of existence of some important excitations of the to calculate fatigue life. It was also considered that the blade
natural frequencies of the blade the important stresses will be surface was altered by erosion and corrosion action (Figure 7)
generated in this zone which can lead to the fatigue failure. which reduce fatigue life in the failure zone using software
nCode [16]. The endurance limit of blade material is around
Fatigue analysis of the blade 440 MPa for the case of smooth and polished surface (which is
Fatigue analysis in the zone of L-0 blade failure (trailing very high value compared to the small alternating stress).
edge) considers excitation of blade fourth mode of vibration by However it can be decreased up to 137 MPa or less in case of
ninth harmonic of rotor speed. The dynamic response of the corrosion or metal surface degradation. The results obtained
blade was calculated from steady state conditions (action of from nCode predict the life of the blade beyond cutoff (infinity
centrifugal force and steam flow pressure as mean stress) and life). It means that the excitation of L-0 blade fourth mode of
from alternating sinusoidal steam flow pressure fluctuation vibration by ninth harmonic of rotor speed could not lead to
during blade row rotation within the range  10% [14].

6 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 blade fatigue failure during turbine operation in steady state to HCF with very important contribution of erosion and
normal conditions. corrosion. In all 37 fractured blades the fracture initiates at the
same zone; zone of changing of section of blade where begins
Metallurgical investigation of the L-0 blade transition radius airfoil-root platform of blade. Due to abrupt
The metallurgical investigation of the failed L-0 blade included change of blade geometry in this zone it is generated stress
metallography, SEM (scanning electronic microscopy) concentration. The maximum static stresses were registered in
fractography and chemical analysis. The microstructure of the the blade trailing edge (zone of failure) reaching value of 569.2
blade consists of tempered homogenous martensite typical for MPa (von Mises) or 574 MPa Maximum principal stress which
forged stainless steel according to specification AISI 420. give moderate safety factor of 1.33 to 1.34 related to blade
The chemical composition and hardness tests of the failed material yield strength (762 MPa). It means that exist small
blades were carried out which confirmed concordance of the margin to accommodate flow induced stresses (dynamic origin)
blade material used to the design specification. The average which can be generate during turbine operation at out of
hardness of blade material was 22 HRC which fall within the nominal conditions (transitions, flow instabilities), for example
design limits. Fractography evaluation was carried out on the LL/LV operation. The results of metallurgical investigation
exposed crack surface of blade airfoil using scanning electronic revealed that fracture initiation points were localized in erosion
microscopy (SEM) to determine the origin of the fracture. picks which rise local stresses comparing to original not altered
Figure 12 shows the zone of the crack initiation where the surface. Also in the same zone (erosion picks) there were found
transgranular crack initiation and propagation was found which corrosion products (deposits). These findings lead to
is typical of a fatigue failure mechanism. conclusions that the HCF failure initiation and propagation of
Blade trailing edge the blades was highly influenced by erosion-corrosion process.
From fatigue behavior viewpoint, dynamic stresses during
operation are very low and are under the fatigue limit not
causing fatigue failure directly. As it was mentioned before the
fourth mode of blade vibration, which has a frequency of 537.6
Hz is very close to ninth harmonic of the rotor speed (540 Hz).
Considering that some uncertainty in calculation of blade
natural frequencies exists it may mean that blade fourth mode of
vibration is in resonance with ninth harmonic of rotor speed.
Fracture initiation points Mistuning phenomenon of the blades was not considered here.
Erosion picks
If the fatigue initiation and propagation were during continuous
Figure 12. Detail of the zone of the fracture initiation (SEM) of operation under resonance vibratory stresses, the blade failure
the blade #42. would occur practically immediately (after a few hours of
operation). Considering that this value and real blades operation
Blade fracture initiated in erosion picks (Figure 12). The period of 59600 hours it can be concluded that blades dynamic
corrosion deposits of S, Ca, Na, K, Al and Si accumulated in characteristics (resonances) don’t contributed to the blades
erosion picks were found. Deposits of elements of blade failure during operation under nominal load (steady state
material Fe y Cr were also found. The depth of erosion picks operation). This conclusion is in a good agreement with results
varied within the range 600 - 700 m and width 200 – 500 m. of blade fatigue life prediction applying nCode presented
The fracture surface was divided a three zones; fracture previously. This conclusion also indicates that fatigue failure of
initiation zone, first stage of fracture propagation and second the blades was not originated during continuous operation under
stage of fracture propagation. The fracture initiation zone and vibration stresses, but during transition events. Analyzing the
zone of first stage fracture propagation represent one event of unit operation history during the year 2005 and 2006, there
excessive stresses (sudden fracture) developed during some were found some transient events which may be related to the
period of operation. The second zone of fracture propagation is blades failure. In total seven periods of operation with low load
characterized by a presence of a several beach marks which are within the range 15 to 20 MW were registered (53 % to 70 % of
characteristic for fatigue mechanism of fracture propagation. nominal load). One example of turbine low load operation
period is shown in Figure 13. As was mentioned previously,
turbine operation with LL can excite a torsional vibration mode
 . of the L-0 blades (X-shaped mode) by unstable flow,
developing high vibratory stresses which can lead to blades
Blade failure analysis failures. Figure 6 shows steam flow stream lines distribution at
Analyzing details of the features registered in the L-0 blades the L-0 stage during low load operation. The typical vibration
(stage 9) presented previously it may be concluded that the time history of the blades in X-shaped mode (Fig.14), which
origin/initiation and propagation of the failure can be attributed corresponds to the fundamental vibration mode, is shown in
Figure 15 [19].

7 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Vibration stress
Generation of Energy 1

0.8

0.6

Normalized vibration stress

0.4

Max. generation 0.2

Max. gen. with reserve
0
MW

Contracted generation
Tech. min. generation -0.2
Real generation
-0.4

-0.6

-0.8

-1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time (sec)
Hours
Figure 15. Vibration time history of the blades in X-shaped
Figure 13. Daily plot of generation, September 14, 2005. mode.

2.3 Failure analysis of last but one (L-1) turbine blade Blade failure analysis
of a 35 MW geothermal unit. Analyzing details of the features registered in the L-1 blades
Background presented previously it may be concluded that the
The blade under evaluation was the 15-inch/3600 rpm origin/initiation and propagation of the failure can be attributed
L-1 blade of a 35 MW geothermal turbine. The blade was made to HCF with contribution of corrosion.
of AISI 410 stainless steel. This unit has double flow low-
pressure turbine composed by 7 stages.

Instant fracture
Fatigue zone zone

Damaged L-0
blades

Figure 14. The X-mode shape of blades vibration.

The L-1 row consists of 74 free-standing blades. The turbine

accumulates of 35000 operation hours to failure. The 7 L-1
blades of turbine side were separated in the root zone causing
related damage of diaphragms, moving blades of the same stage
and L-0 blades of the turbine. The non destructive tests carried
out, revealed that 5 L-1 blades of generator side were cracked at
the same zone but without airfoil separating. Also, turbine
casing was damaged. Figure 16 shows damage registered in the Figure 16. Damage registered in the turbine.
turbine which includes detail of damaged L-1 blades, general a). Detail of the L-1 blade fracture, b). Damaged L-1 blades, c).
aspect of L-1 blade fracture surface, damaged L-0 diaphragm Casing damage.
and turbine casing damage. Fractography evaluation was carried
out on the exposed crack surface of blade root using scanning
electronic microscopy (SEM) to determine the origin of the
fracture. Transgranular crack initiation and propagation was
found which is typical of a fatigue failure mechanism. 3744 rpm (4 % above nominal speed of
3600 rpm).

8 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 compound low-pressure turbines. Each L-0 row consists of 180
blades in 45 groups of 4, with the 4-blades in each group
connected by 3 tiewires. During the unit’s last overhauls, the
turbines visual examination revealed that in the unit 1, LP1
rotor, 8 L-0 blades on the governor side (4.5 %) and 23 blades
on the generator side (13 %) were cracked, and also in LP2
rotor, 37 blades on the governor side (21 %) and 18 blades on
the generator side (10 %) were cracked too. Similarly in the unit
2, LP1 rotor, 12 L-0 blades on the generator side (7 %) were
cracked with no-cracked blades on the governor side, and also
As it can be in LP2 rotor, 21 blades on the governor side (12 %) and 8
seen in Figure 16a, the fatigue zone of the fracture surface is blades on the generator side (4.5 %) were cracked too. The
divided by beach marks, which are characteristic for fatigue blade is made of 17-4 PH precipitation hardening stainless steel.
mechanism of fracture propagation. The number of beach marks The failed blades had cracks in their roots initiating at the
on the failed blades fracture surface was variable from blade to trailing edge, concave side of the steeple outermost fillet radius
blade. The largest number of beach marks found in the blade as is shown in Fig. 17. All the cracks initiated from blade root
No. 26 was 15 (Figure 16a). Analyzing turbine operation serrations, at the corner under the trailing edge where the
history, it was found that since turbine commissioning date, it concave side of the root meets the exhaust side endface (Fig.
experienced 18 emergency trips due to vacuum pump problems 18). The detail of the fracture surface is shown in Fig. 18b with
which operate in unit non condensable gases removal system. various features identified by metallographic examination.
Instant shut down of vacuum pump typically cause sudden
change of turbine/condenser vacuum (it is generated counter
pressure), before turbine is tripped and it represent strong
excitation of low pressure blades similarly as it is presented in Fracture

Figure 6. It is worth to mention that in this case the main

vacuum system was not equipped in auxiliary/emergency
vacuum system which can be operated immediately if main
vacuum system failed. T

Fracture

Figure 17. Fracture in the L-0 blade root (unit 2).

As it can be seen, the fracture surface is divided by a beach

marks (BM) in the zones of different roughness where the
It was concluded presence of striations-fracture sliding planes (S), which are
that high cycle fatigue failure of L-1 blades were caused characteristic for fatigue mechanism of fracture propagation, are
probably by high excitation of blades during vacuum pump trips noticeable. The number of beach marks on the failed blades
and turbine operation periods with low load. fracture surface was variable. The smallest number of beach
2.4 Failure analysis of last stage (L-0) turbine blade marks found in unit 2 was 3 and the highest was 13 (see Fig.
of a 660MW nuclear unit. 18a). The smallest distance between the beach marks was 0.26
Background mm and the biggest was 12.2 mm. The longitude of the fracture
The blade under evaluation was the 44-inch/1800 rpm last in all failed blades was varied within the range 11.6 mm to
stage blade (L-0) of a 660 MW turbine (two units) which 19.02 mm.
consists of one high-pressure turbine and two tandem-

9 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Blades natural frequency analysis
The metallurgical investigation of the failed L-0 blade of To assess whether the dynamic response characteristics of
unit 2 was carried out, and included metallography, SEM the L-0 bladed-disk row could have contributed in any way to
(scanning electronic microscopy) fractography and chemical the observed failure, the zero speed natural frequency tests were
analysis. The microstructure of the blade consists of tempered carried out. These included a rowing impact-hammer modal test
homogenous martensite obtained due to precipitating hardening of all 4-blade groups in all L-0 rows of unit 2. The results of
heat treatment, free of failures, typical for forged stainless steel these 0 rpm test were transformed to 1800 rpm rotational speed
according to specification 17-4 PH (AISI 630). The chemical and are presented in Figure 20 (LP2 rotor, high-pressure side)
composition and hardness tests of the failed blades were carried as a representative for the all L-0 stages. The vibration response
out which confirmed concordance of the blade material used to (amplitude) of the groups of blades is expressed in g’s RMS
the design specification. The average hardness of blade material normalized dimensionless. As it can be seen from Fig. 19 there
was 26.5 HRC which fall within the design limits. Fractography is some frequency scatter between groups of blades particularly
evaluation was carried out on the exposed crack surface of for the second mode which can be explained by variation in
blade root using scanning electronic microscopy (SEM) to erosion rate between blade to blade, variation of the blades
determine the origin of the fracture. Figure 19 shows the zone geometrical tolerances of manufacturing and blade group’s
of the crack initiation where the transgranular crack initiation installation tolerances variation. The calculated natural
and propagation was found which is typical of a fatigue failure frequencies of the groups of the blades (see Fig. 20) confirm
mechanism. that no mechanical resonances of the blade structure exist.
There is no coincidence of any harmonic of rotor rotating speed
and blade natural frequency.
LP2- high pressure side

Amplitude 0.8

0.6

Crack initiation point 0.4

a)
0.2
S
BM 0
30 60 90 120 150

Frequency-Hz

S
Figure 20. The 1800 rpm dynamic response of the first three
modal forms of vibration of the L-0 blades of the LP2 rotor,
S
high-pressure side, unit 2.
S
Blade failure analysis
Crack initiation point Considering the results of the L-0 blades natural frequency
b)
analysis and blades real operation period (nine years=70,000
Figure 18. Detail of the L-0 blade fracture surface (unit 2). BM hours of operation approximately), it may be concluded that the
- beach marks, S – striations. mechanical resonance of the blades does not contribute to blade
failure. If the fatigue initiation and propagation were during
continuous operation under resonance vibratory stresses, the
blade failure would occur practically immediately (after a few
hours of operation) as it is explained as follows. The time t to
fatigue fracture initiation can be determined from equation (1):
N
t  [s] (1)
f
Introducing to equation (1) value of f = 30 Hz and 120 Hz (first
and fourth harmonic; see Fig. 19), and N = 107 Hz; fatigue life
of blade material, will give the life of the blade under resonance
t = 92 to 23 hours respectively. Consider this value and real
blades operation period mentioned before it can be concluded
that blades dynamic characteristics (resonances) don’t
contributed to the blades failure during operation under nominal
Figure 19. Detail of the zone of the fracture initiation (SEM). load (steady state operation). This was also plotted in a

10 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Campbell diagram including up to the first ten frequencies and grouped blades which results in high quantity of cycles of high
all harmonics nearby, but not shown here for space reasons. stresses specifically in L-0 blades due to their major length and
Analyzing the unit 1 and unit 2 operation history since the date weight. Accumulation of these cycles can lead to the cracks
in which the L-0 blades were installed, there were found some initiation and propagation due to fatigue. It is also facilitate
transient events which may be related to the blades failure. In because of significant length of the turbo generator rotor (50 m
the case of unit 1, one period of operation with low load/low approximately) which represent high torsional inertia. Figure 23
vacuum was registered (50 minutes) and 13 events in total of (a) provides the frequency spectrum of the power time history
operation with poor vacuum and reduced load (vacuum below of similar units with “low excitation” while Figure 23 (b)
the design value but within the operation range limits) and provides the “high excitation” [15]. Note the 114.75 Hz
sudden load changing. In the case of unit 2 there were addition to the signal with the 60 and 120 Hz background noise.
registered in total 30 events of sudden load changing and Vibrations

operation periods with poor vacuum/reduced load. Some part of 0.12

SLC
BEARING 9 HORIZ.

these events may be related to fracture initiation (incubation)

BEARING 9 VERT.

0.1

period and other part to fracture propagation period. During

Displacement p-p (mm)

unit’s last overhaul, it was carried out non destructive 0.08

examination of all L-0 blades and there was not found any 0.06

cracks indications in the blades roots. Since this time, during

current inspection which revealed blades failure, it was found 0.04

maximum 13 beach marks on one blade fracture surface of unit 0.02

2 (Fig. 18a), which probably corresponds to all fatigue

propagation events registered in this period. On the other hand, 0

03-Apr-02 00:00
03-Apr-02 15:00

04-Apr-02 21:00
05-Apr-02 12:00
06-Apr-02 03:00
06-Apr-02 18:00

08-Apr-02 00:00
08-Apr-02 15:00

09-Apr-02 21:00
10-Apr-02 12:00
11-Apr-02 03:00
11-Apr-02 18:00
12-Apr-02 09:00
13-Apr-02 00:00

14-Apr-02 06:00
14-Apr-02 21:00

16-Apr-02 03:00

18-Apr-02 00:00
16-Apr-02 18:00

18-Apr-02 15:00
19-Apr-02 06:00

20-Apr-02 12:00
21-Apr-02 03:00

22-Apr-02 09:00
23-Apr-02 00:00
23-Apr-02 15:00
04-Apr-02 06:00

07-Apr-02 09:00

09-Apr-02 06:00

13-Apr-02 15:00

15-Apr-02 12:00

17-Apr-02 09:00

19-Apr-02 21:00

21-Apr-02 18:00
at the same period since last overhaul, there were registered 14
events in unit 2, which may be related to blade fracture
propagation period (3 events of sudden load changing and 11 time

events of operation with poor vacuum/reduced load). Figure 22. Unit vibration increase due to sudden load changes
Comparing these data, it is observed a very good agreement in the grid (SLC).
between number of beach marks on the blade fracture surface
and number of hypothetic fatigue fracture propagation events. Figure 24 provides a time history of the torsional responses of
In Fig. 21 it is shown an example of one transient event of the rotor. Notice the change in torsional amplitude which varies
sudden load changing in electrical grid due to failure of with power. Figure 25 provides a frequency spectrum of the
neighborhood electric line (load drop of 74 MW) related to unit time history. Note the coincidence at 114 Hz; the perturbation in
2 and in Fig. 22 vibration increase of this unit due to this event. power at 114.5 Hz is driving the vibration response at 114.26
From reference [15] it is known that the analyzed train of rotors Hz. According to references [6, 14 and 19] turbine operation
of the 660 MW units is susceptible to sporadic/transient with low vacuum/low load can excite a torsional vibration mode
torsional vibration. As it was mentioned before, two types of of the L-0 blades (X-shaped mode) by unstable flow,
changes in the turbo generator (t/g) system can cause torsional developing high vibratory stresses. As it was mentioned
vibration. previously, these operation conditions generate blade excitation
Load vs ti me
forces (X mode torsional vibrations), which can lead to blade
failures. The typical vibration spectrum of the grouped blades in
700
X-shaped mode, which corresponds to the fundamental
650
Load vs t ime vibration mode, is shown in Figure 15 [19]. The
600
characterization of stresses existing at the blades during
550
operation for 1sth tangential mode, 3rd X-shaped mode and for
MWe

500
mixed mode (1st tangential mode and 3rd X-shaped mode
450
SLC together) of vibration in group of 4 blades is shown in Figure
400
26. As it can be seen from Fig. 26, for 3rd mode of vibration (X-
350
shaped) the highest stresses are developed at the 1st blade in the
300
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
group, next at the 4th and 2nd, and finely at 3rd blade. For the 1st
Tim e (April 2002) [days] vibration mode (tangential) the highest stresses are at the 4th
blade in the group (last blade of the group), next at the 3 rd, 2nd
Figure 21. Transient event of sudden load changes in electrical and 1st blade. For the 3rd mode of vibration it can be observed
grid (SLC). large differences in stress levels between first and the rest of the
blades in the group (especially third blade). According to this
These transition events can excite torsional resonance of the stress distribution at the blades the frequency of the blades
rotor resulting in exciting of 1st mode of vibration (tangential) failure at each position in the group should have similar

11 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 distribution. Failure of the "end" blades of a group could be also with the 1st tangential mode, depending on stages (rotors)
consistent with an "X-shaped" mode of vibration. In the case of analyzed.
1st mode of vibration, the stress distribution at the blades is
0.048

more uniform and the frequency of the blades failure at each 0.040

position in the group should have more random distribution, i.e.

RAD/SEC RMS
0.032

anywhere within the group. 0.024

4.8

0.016
4.0

0.008
3.2
MWATTS RMS

0.000
0 40 80 120 160 200
FREQUENCY IN HZ
AMP: 0.006174 FREQ: 114.26
2.4

(a)
1.6

0.048
0.8

0.040
0.0

RAD/SEC RMS
0 40 80 120 160 200 0.032
FREQUENCY IN HZ
AMP: 0.01168 FREQ: 114.75
0.024
(a)
0.016

0.008
4.8

0.000
0 40 80 120 160 200
FREQUENCY IN HZ
4.0

AMP: 0.3117 FREQ: 114.26

(b)
3.2
MWATTS RMS

Figure 25. Frequency spectrum of torsional vibration, (a) Low

excitation and (b) High excitation.
2.4
1.6

100
0.8

90
Normalized stresses in the zone of failure

80
0.0

70
0 40 80 120 160 200
60
FREQUENCY IN HZ
AMP: 3.054 FREQ: 114.75 50

(b)

Figure 23. Frequency spectra of generator power, (a) Low 30

3rd. Mode
excitation and (b) High excitation [15]. 20 1st. Mode
10

0
Low response High response 1 2 3 4

2.00 Blade position in the group

RAD/SEC

0.00

-2.00
2.00 Low excitation High excitation
100
RAD/SEC

0.00 90
Normalized stresses in the zone of failure

80
-2.00
0.000 10.432 38.912 59.392 79.872 90.112 102.400 112.640 70
TIME IN SECONDS
60
Figure 24. Time history of rotor torsional vibration [15].
50
1st and 3rd
40 mixed mode
In the case of mixed mode, there are small differences in stress 30

levels between the blades; and stress distribution is similar as 20

for 3rd mode of vibration. In this case, the frequency of the 10

blades failure at each position in the group should have more 0

1 2 3 4
random distribution than for the 3rd X-shaped mode. Analyzing Blade position in the group

the distribution of the failed blades in the L-0 stages of unit 2

(see Fig. 27) it can be observed that the distribution of the
Figure 26. a). Effect of vibration for each blade for 1st and 3rd
cracked blades is consistent with the 3rd X-shaped mode and
modes; and b). for 1st and 3rd mixed mode.

12 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 vibration due sudden load changing on the grid, the transient
20 torsional vibration monitoring systems installation in the units
18
may be helpful to de-tune rotor torsional mode close to 120 Hz,
16
Number of failed blades

14
implementing modifications of the rotor mass.
12 LP2-HP
LP2-GEN
10

8
LP1-HP
LP1-GEN
4 REFERENCES
[1] Hernández-Rossette A., Mazur Z. and García-Illescas R.,
blades

4
Análisis de la Falla Presentada en la Rueda L-0 del Rotor
2

0
de la Unidad 7 Instalada en la C.G.T. Cerro Prieto II (In
1 2 3

Blade position in the group

4
Spanish). Report IIE/43/12900/I001/F/DI/V1, 2005,
Cuernavaca.
Figure 27. Distribution of the failed blades in the L-0 stages of [2] Mazur Z., García-Illescas R., Porcayo-Calderón J.,
the unit 2. Evaluación de la Causa de la Falla del Álabe Móvil Etapa
9 de la Central Geotérmica Berlín, El Salvador (In
This distribution is influenced by the number and duration of Spanish). Report IIE/43/13234/I002/F/DI/A1/V1, 2008,
fatigue events (operation periods with low load low/poor Cuernavaca.
vacuum and events of sudden load changes) and which of these [3] Mazur Z., Evaluación Preliminar de la Falla de los Álabes
two groups of events predominated. Etapa L-1 de la Turbina Fuji de 35 MW, de la Unidad 3,
de la Central Geotérmica Ahuachapán (In Spanish).
3 CONCLUSIONS Report IIE/43/I/3162/005/P, 2007, Cuernavaca.
Flow-induced blade oscillations of the turbine can lead to [4] Mazur Z., García-Illescas R., Aguirre-Romano J. and
blade fatigue failures and so they represent an important Pérez-Rodríguez N., Apoyo para Determinar las Posibles
problem of reliability, safety, and operating cost. Aeroelasticity Causas Raíz de la Falla de los Álabes de las Turbinas de
phenomena are characterized by the interaction of fluid and Baja Presión de las dos Unidades de la CLV (In Spanish),
structural domains. Aeroelasticity phenomena can occur while Report IIE/43/2827/I/007/P, 2004, Cuernavaca.
turbine is operating at LL/LV. During turbine operation at a [5] Coffer, J.I. et al, Advances in Steam Path Technology, GE
LL/LV, the low pressure blade vibration stresses are increasing Power Systems, GER-3713E, New York, 1996.
abruptly. This increase (peak) of the blade vibration is induced [6] Suzuki T., et al., Recent Upgrading and Life Extension
by unstable flow with oscillation of a shock wave near the Technologies for Existing Steam Turbines, ASME Power,
throat of the blade tip passage. Recently, a remarkable progress April 5-7, 2005, Chicago, pp.577-582.
in transient flow calculations allowed the prediction of more [7] Szwedowicz J., et al, On Forced Vibration of Shrouded
realistic excitation forces acting on the rotating blades. To Turbine Blades, Proceedings of ASME Turbo Expo 2003,
produce this work the transient flow calculation not has been Power for Land, Sea, and Air, June 16-19, 2003, Atlanta,
carried out. Nevertheless, due to some current uncertainties in Georgia, USA.
transient flow calculation and forced response of blades [8] Filsinger, D., et al, January 2002, Approach to
assemblies, as was reported herein, some cases of steam turbine Unidirectional Coupled CFD-FEM Analysis of Axial
low pressure blades failures structure have been recorded. Other Turbocharger Turbine Blades, Transaction of the ASME,
problem which exists in turbine and generator (t/g) rotor Journal of Turbomachinery, Vol. 124, pp. 125-131.
components are torsionally induced high cycle vibration. This is [9] Kielb. R. et al, Blade Excitation by Aerodynamic
generally a sporadic, transient phenomenon provoked by Instabilities-a Compressor Blade Study, Proceedings of
sudden load changes on the grid and/or inter-harmonic loading. ASME Turbo Expo 2003, Power for Land, Sea, and Air,
Most of the times these transient events do not overly excite the June 16-19, 2003, Atlanta, Georgia, USA.
t/g shaft torsional resonances, but in some cases there is a [10] Kielb R., et al, Flutter of Low Pressure Turbine Blades with
coincidence of the transient’s wave form characteristics and Cyclic Symmetric Modes-a Preliminary Design Method,
torsional resonances resulting in several cycles of high stress. Proceedings of ASME Turbo Expo 2003, Power for Land,
The accumulation of these cycles may lead to crack initiation Sea, and Air, June 16-19, 2003, Atlanta, Georgia, USA.
and fatigue failure. To reduce the probability of steam turbine [11] Silkowski, P.D. et al, CFD Investigation of Aeromechanics,
low-pressure blades failure, the operational conditions should Proceedings of ASME Turbo Expo 2001, Power for Land,
be maintained within the design limits avoiding operation with Sea, and Air, June 4-7, 2001, New Orleans, Louisiana,
low load/low vacuum. In the case of application of vacuum USA.
pumps in the unit non condensable gases removal system, the [12] Panowsky J. an Kielb, R.E., A Design Method to Prevent
main vacuum system should be equipped with Low Pressure Turbine Blade Flutter, Journal of
auxiliary/emergency vacuum system, which can be operated Engineering for Gas Turbines and Power, January 2000,
immediately if main vacuum system fails. In the case of large Vol. 122, pp. 89-98.
nuclear units, to mitigate torsionally induced high cycle

13 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 [13] Vogt, D.M. and Fransson T.H., A New Turbine Cascade for [16] Program nCode, International Limited, Version 1, 2001.
Aeromechanical Testing, The 16th Symposium on [17] Nisa Module, Engineering Mechanics Research
Measuring Techniques in Transonic and Supersonic Flow Corporation, 2000.
in Cascades and Turbomachines, Cambridge, September [18] Abdulla S. Rangwala, Book: "Turbomachinery Dynamics,
2002, UK. Design and Operation", Mc Graw Hill, 2005.
[14] Troyanowskij B.M., Filipow G.A. and Bulkin A.E., [19] Report of L-0 Rotating Blades Failure, T8-04109,
Parovyje i Gazovyje Turbiny Atomnych Elektrostancyj, (in Mitsubishi Heavy Industries, Ltd., May, 2004.
Russian), Moscow, Energoatomizdat, 1985. [20] Mazur Z., Salazar San Andrés O. R., Metodología para
[15] Dorfman L.S. and Trubelja M., Torsional Monitoring of Determinar la Longitud de Álabes del Último Paso de las
Turbine-Generators for Incipient Failure Detection, Sixth Turbinas de Vapor (In Spanish), Report
EPRI Steam Turbine/Generator Workshop, August 17-20, IIE/43/12618/I001/F, 2004, Cuernavaca.
1999, St. Luis, Missouri, pp.1-6.

14 Copyright © 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use