Available online at www.sciencedirect.

com

Engineering Failure Analysis 15 (2008) 1111–1129

www.elsevier.com/locate/engfailanal

Failure analysis of a second stage blade in a gas turbine engine

E. Poursaeidi a,b,*, M. Aieneravaie b, M.R. Mohammadi b
a
Department of Mechanical Engineering, Zanjan University, Zanjan, Iran
b
IPPR Research Lab, Iran Power Plant Repairs Co, P.O. Box 31585-381, Karaj, Iran

Received 6 July 2007; accepted 24 November 2007

Available online 14 December 2007

Abstract

The failure of a second stage blade in a gas turbine was investigated by metallurgical and mechanical examinations of
the failed blade. The blade was made of a nickel-base alloy Inconel 738LC. The turbine engine has been in service for about
73,500 h before the blade failure at 5:50 PM on 14 August 2004. Due to the blade failure, the turbine engine was damaged
severely. The investigation was started with a thorough visual inspection of the turbine and the blades surfaces, followed by
the fractography of the fracture surfaces, microstructural investigations, chemical analysis and hardness measurement.
The observation showed that a serious pitting was occurred on the blade surfaces and there were evidences of fatigue
marks in the fracture surface. The microstructural changes were not critical. It was found that the crack initiated by the hot
corrosion from the leading edge and propagated by fatigue and finally, as a result of the reduction in cross-section area,
fracture was completed.
An analytical calculation parallel to the finite element method was utilized to determine the static stresses due to huge
centrifugal force. The dynamic characteristics of the turbine blade were evaluated by the finite element modal and har-
monic analyses. Finally according to the log sheet records and by using a Campbell diagram there was a good agreement
between the failure signs and FEM results which showed the broken blade has been resonated by the third vibrational
mode occasionally before the failure occurred.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Blade failure; Hot corrosion; Fatigue failure; Inconel 738LC; Finite element

1. Introduction

Superalloys were developed since the second quarter of the 20th century as materials for elevated temper-
ature applications and can be divided in three groups: nickel-base superalloys, cobalt-base superalloys and
iron base superalloys. Gas turbine blades are principally made of nickel-base and cobalt-base superalloys.
The main reason for the existence of superalloys is their outstanding strength at elevated temperatures, which
make them suitable for the fabrication of gas turbine components. During the operation of power generation

*
Corresponding author. Address: Department of Mechanical Engineering, Zanjan University, Zanjan, Iran. Tel.: +98 912 2133496; fax:
+98 261 6605929.
E-mail addresses: epsaeidi@yahoo.com, epsaeidi@znu.ac.ir (E. Poursaeidi).

1350-6307/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2007.11.020
1112 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

gas turbines, the blades and other elements of hot gas path undergo service-induced degradation, which may
be natural or accelerated due to different causes [1,2].
The degradation or damage may have a metallurgical or mechanical origin and results in reduction of
equipment reliability and availability. To identify the causes of the blade failures, a complete investigation
has to be carried out, integrating both the mechanical analyses and metallurgical examination. Metallurgical
examination can be very effective in determining whether the failure is related to material defects, mechanical
marks, poor surface finish, initial flaws or heat treatment.
There are different factors, which influence blade lifetime, as design and operation conditions but the latter
are more critical. In general, most blades have severe operation conditions characterized by the following fac-
tors [3]:

 Operation environment (high temperature, fuel and air contamination, solid particles, etc.).
 High mechanical stresses (due to centrifugal force, vibratory and flexural stresses, etc.).
 High thermal stresses (due to thermal gradients).

Typically there are acting two or more factors simultaneously causing reduction of blade lifetime. The type
of damage, which occurs in gas turbine blades and nozzles after a service period, can be divided into [4–7]:

 External and internal surface damages (corrosion, oxidation, crack formation, erosion, foreign object dam-
age and fretting).
 Internal damage of microstructure, such as c0 [Ni3(Al,Ti)] phase aging (rafting), grain growth, creep and
grain boundary void formation, carbides precipitation and brittle phases formation.

Surface damage produces blades/nozzles dimensional changes, which result in operational stress increase
and turbine efficiency deterioration.
In service, blade material deterioration is related to the high gas temperature, high steady state load levels
(centrifugal load) and high thermal transient loads (trips, start-ups and slowing downs). However, the degree
of deterioration in individual blades differs due to several factors such as:

 Total service time and operation history (number of start-ups, shut-downs and trips).
 Engine operational conditions (temperature, rotational speed, mode of operation (base load, cyclic duty)).
 Manufacturing differences (grain size, porosity, alloy composition, heat treatment).

The Inconel 738LC alloy commonly used for gas turbine blades is strengthened by precipitation of c0 phase.
The microstructural changes due to blade operation at high temperature include irregular growing of particles
(rafting) and formation of carbides in grain boundaries and matrix [2]. This leads to alloy creep properties
reduction.
In order to have an instrument for the deterioration evaluation of gas turbine blade alloy, it is necessary to
associate the influence of service-induced microstructural degradation to the changes in mechanical properties.
This can be used for monitoring and evaluation of extent and degree of material damage and lifetime con-
sumed and to obtain recommendations for blade rejuvenation treatments, operation and reposition [8,9].
Application of effective methods of material deterioration evaluation can be used for practical lifetime pre-
diction, just in-time blade rehabilitation (rejuvenation), safe and cost-effective lifetime extension and to avoid
blade catastrophic failure.
In the other hand most of gas turbo generators are used as an auxiliary compensator in power plants to
generate electric power at the peak of load demand. Thus they often are utilized in discontinuous conditions
of commissioning. This subject leads them to a lot of shocks and risks. Therefore it seems the failure analysis is
a good manner for detecting the root causes. So according to the results of failure analysis, the gas turbine
would be utilized by applying some new policies in the protected conditions. Often by using an intelligent
mechanical analysis, the root causes of a failure could be revealed. Recently the computer programs and soft-
ware packages are generalized to calculate the mechanical behavior of gas turbine blades. In this paper the
ANSYS code was applied for generating and simulating a FE model of fractured blade.
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1113

2. Experimental procedure

The blade material was a cast Ni-base superalloy known as Inconel 738LC with the mechanical specifica-
tions that is presented in Table 1. Because of relation between the working temperature of alloys and their
specifications, the tensile yield strength of all parts at the working temperatures were considered along this
case study. According to the thermal conditions at the peak load of these types of gas turbine engine, the tem-
perature of combustion products around the surfaces of airfoil arises up to 560 °C on the leading edge and
520 °C on the trailing edge (Fig. 1).
The turbine has been in service for about 73,500 h and due to the blade failure (at 5:50 PM on 14 August
2004) the turbine engine was damaged severely. In addition to the failed blade, two adjacent blades were sub-
jected to the laboratory for extensive failure analysis. The following sequence of examination was performed
on the blades:

1. Visual examination and photographic documentation.

2. Hardness measurement of the transverse section beneath the fracture surface.
3. Chemical analysis of the blade material.
4. Metallographic examination of the transverse section of samples from the airfoil and root blade, and micro-
probe analysis.
5. Examination of the fracture surface by means of optical and electron microscopes.

3. Experimental results

3.1. Visual examination

Fig. 2a and b shows the failed blade in the second stage of turbine, which signed as number 1. As a result of
failure the fragmented part has damaged the leading edges and shroud tips of some blades (Fig. 2b). Three
blades (which signed in Fig. 2) were submitted to the laboratory. The tip regions of the blades 2 and 3 were
severely damaged presumably by impact; however the tip damage was not the focus of this investigation. All
three blades were corroded by the pitting mechanism. The region of pitting on the concave surface of the air-
foil was near the leading edge, and on the convex surface was near the trailing edge (Figs. 3a and b). The pits
on the leading edge were more critical and in some cases appeared like the transverse notches (Fig. 4). It
should be noted that the height of the fracture surface from the platform at leading edge in blade number
1 was the same as the position of critical pits in blade number 3 (Fig. 5). There was not a considerable deposit
on the blades surfaces.

3.2. Chemical analysis

To determine the chemical composition of the blade material a bulk analysis was performed by an Optical
Emission Analyzer, ARC-MET 930 S&P. The result of chemical analysis showing that the material is in accor-
dance with Inconel 738LC (Table 2).

3.3. Hardness measurement

The hardness of the transverse section of the blade beneath the fracture surface was measured by a Univer-
sal Hardness Measurement Machine, SWISS MAX 300. Fig. 6 shows the points of hardness measurement on

Table 1
Material properties of alloys at the ambient temperature [10–12]
Component Material Mass density (kg/m3) Tensile yield strength [Min] (Pa) Tensile ultimate strength (Pa)
8
Bucket IN-738LC 8110.0 9.5  10 1.1  109
Disc A471-Cl.10 7800.0 6.2  108 7.25  108
1114 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

Fig. 1. The sketch view of turbine’s hot section with temperatures of some important points (presented by the manufacturer).

Fig. 2. Position of the failed blade and the damages were caused due to failure. (a) Upper view of the failed blade and surroundings.
(b) Leading edge damage in some blades.
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1115

Fig. 3. General view of the blade number 3 showing corroded regions. (a) Concave surface. (b) Convex surface.

the blade cross-section. The results are given in Table 3. It is observed that there was no substantial change in
the material’s hardness, and thermal effects did not influence the strength of the material.

3.4. Metallographic examination and microprobe analysis

Samples for microstructural evaluation were taken from the blade airfoil near the fracture surface and
blade root. Fig. 7 illustrates the micrographs of the material. As to be observed the microstructure has com-
posed of dendritic grains (Fig. 7a), which is a characteristic of cast structures. There is a relatively compact
scale on the surface of the blade. Although this scale was not continuously built up all over the surface, in
some regions there is a noticeably thick scale (Figs. 7b and c). Underneath the scale, a severe deterioration
1116 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

Fig. 4. The critical pit which is encircled in Fig. 3a. 20.

Fig. 5. The distance of the fracture surface from the platform at blade edges.

can be noticed. The severity of this deterioration was to a greater degree in the samples of leading edge, espe-
cially on the concave surface. This is in relation with the results of visual examination. Figs. 7a and b shows
that there are a lot of microcracks where initiated from the surface. Metallographic examination at higher
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1117

Table 2
Chemical composition of Inconel 738LC superalloy (wt%)
Alloy Ni C Cr Co Mo W Al Ti B
IN 738LC Bal 0.11 16 8.5 1.75 2.6 3.4 3.4 0.01

Fig. 6. Points of hardness measurement on the blade cross-section.

Table 3
Hardness of the transverse section of the blade
Point of indentation 1 2 3 4 5 6 7 8 9 10 11
Hardness (HV30) 410 408 400 395 395 396 389 387 406 410 417

magnifications carried out by means of scanning electron microscope (SEM). The distribution of
c0 ([Ni3(Al,Ti)]) particles and rarely precipitation of interdendritic carbides can be seen in Fig. 8a. It is observed
from Fig. 8b that the original cubical shape of c0 has not changed.
Microprobe analysis was carried out on the scale and subscale regions (Fig. 9) using an Energy Dispersive
X-ray (EDX) analyzer attached to the SEM. The surface scale is comprised of three main components marked
with number 1 to 3. In the external region (number 1), the scale composed of relatively unprotective nickel and
cobalt oxides. The middle area (number 2) is primarily constituted of protective chromium and aluminum oxi-
des. The internal layer (number 3) is comprised of chromium sulfide and chromium oxide which is the sign of
hot corrosion. Furthermore, presence of sulfur at the subscale (number 4) was detected. Also it is observed
that the nickel percentage was decreased and the cobalt percentage was increased at subscale which is signi-
fying the depletion of nickel from subscale.

3.5. Analysis of the fracture surface

Fracture surface of the failed blade was examined by stereomicroscope and SEM. General view of the frac-
ture surface is shown in Fig. 10. Two regions were clear on the surface. The first region which has marked with
a rectangular (in Fig. 10) is the crack initiation area and the second region which has marked with an ellipse is
the crack propagation by fatigue mechanism. Fig. 11 has taken from the area inside the rectangular of Fig. 10
at higher magnification. As observed in this figure the dendritic morphology is clear on the surface. This mor-
phology demonstrates that the crack path has been interdendritic at initiation area. It can be considered as a
sign of hot corrosion. Fig. 12 exhibits a high magnification fractograph from the region inside the ellipse in
Fig. 10. In this area fatigue marks (striations) are evident. It indicates that after the initiation of the crack,
propagation has conducted by the fatigue mechanism due to the vibration of the blade. Fig. 12 illustrates
two fractographs of different magnifications from disconnected regions in the vicinity of fatigue crack prop-
agation. It demonstrates that the crack growth was occurred with different stress levels and frequencies. After
crack growth and subsequent reduction in cross-section area, final fracture has occurred in the trailing edge of
the blade.
1118 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

Fig. 7. Optical micrograph of the blade material. (a) Dendritic microstructure. (b) Surface scale and subscale deterioration. (c) Same as (b)
from another place.

4. Mechanical analysis

4.1. Analytical calculations

4.1.1. Centrifugal forces

One of the most unfavored loads in the rotating components of gas turbine is the centrifugal force, which
generates tensile stresses along the blade length in the radial direction. It depends on two variable parameters:
the whirling speed of rotor and the distance of each position from the rotating axis.
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1119

Fig. 8. SEM micrograph of the blade material. (a) The particulate c0 and grain boundary carbides. (b) The shape of c0 phase.

Fig. 9. Regions of microprobe analysis.

Hence by applying a nominal rotating speed about 5100 rpm (85 rps); the centrifugal force was simulated
on the elements of FE geometric model and so the values of related stresses. By attention to Fig. 13, the value
of tensile stress r(r) due to centrifugal force F(r) can be estimated from the following equation [10]:
1120 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

Fig. 10. General view of the fracture surface.

Fig. 11. Dendritic morphology in the vicinity of the rectangular in Fig. 10. 20.

rðrÞ ¼ F ðrÞ =S ðrÞ ; ð1Þ

F ðrÞ ¼ F blðrÞ þ F b ; ð2Þ

where Fb is the centrifugal force due to the banding, Fbl(r) the centrifugal force due to the portion of the blade
confined between the sections r  Rh and Rt. It is then clear that
F b ¼ mb x2 ðRTip þ d=2Þ; ð3Þ
where mb is the banding mass, x the angular velocity of rotor, RTip the tip radius, d the banding thickness.
By taking an infinitesimal element dn be separated in section n, the force Fbl(r) can be found as:
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1121

Fig. 12. Fatigue striations in the vicinity of the ellips in Fig. 10 from different regions. (a) 4000. (b) 8000.

Fig. 13. A moving blade that is detailed for selecting a radial element.

Z l
F blðrÞ ¼ qx2 S ðnÞ ðRHub þ nÞdn; ð4Þ
rRHub

where l = RTip  RHub is the length of blade, and S(n) is the function of airfoil cross-section area.
1122 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

As follows from equations (1) and (3), the stresses from centrifugal forces in a blade of arbitrary profile can
be determined by the relationship:
 Z l  
x2 d
rðrÞ ¼ q S ðnÞ ðRHub þ nÞdn þ mb RTip þ : ð5Þ
S ðrÞ rRHub 2
In this section as a predictive assessment by using the above equations, the tensile stress of a single bucket
has been calculated. The real values of required parameters are; q = 8110 kg/m3, d = 0.003 m, mb = 0.12 kg,
RHub = 0.588 m, RTip = 0.808 m, and the omega (x) can be calculated as follow:
2pn 2p  5100
x¼ ¼ ¼ 534 Rad=s: ð6Þ
60 60
Using the equation (3), the centrifugal force of banding portion is calculated:
F b ¼ ð0:12Þð5342 Þð0:808 þ 0:003=2Þ ¼ 27700 N:
The function of airfoil section area was calculated by an interpolation method:
S ðfÞ ¼ ð3:2568f3 þ 1:9592f2  0:5923f þ 0:1032ÞE  02 m2 : ð7Þ
By substituting S(f), Fbl and F(r) are calculated:
F ðrÞ ¼ F b þ F bl
¼ 27700 þ ð1778542:03329 þ 15078250:84320r5  44579357:39326r4 þ 48382501:74297r3
 20667009:12866r2 Þ N:
Finally by substituting the calculated functions in equation (1) the function of tensile stress will be result as
follow:
rðrÞ ¼ ð27700 þ ðð1778542:03329 þ 15078250:84320r5  44579357:39326r4 þ 48382501:74297r3
3 2
 20667009:12866r2 ÞÞÞðð0:0326ðr  0:588Þ þ 0:0196ðr  0:588Þ  0:0059ðr  0:588Þ
1
þ 0:0010ÞÞ Pa: ð8Þ
The plot of centrifugal stress versus rotating radius is shown in the Fig. 14. By attention to the position of
fracture on the broken blade, the stress value is about 186 MPa.

4.1.2. Aerodynamic forces

The aerodynamic loads on the surfaces of the moving blades are calculated by using the turbo-machine the-
ory. By attention to Fig. 15 and considering a control volume around a moving blade in a turbine cascade and

Fig. 14. Graph of tensile stress distributed along the blade.

 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1123

Fig. 15. Scheme for deriving the equation of momentum for a turbine stage.

using the equation of momentum in two directions, the aerodynamic load distribution along the blade length
can be calculated [10]:

qu ¼ qb  C 2a  ðC 1u þ C 2u Þ  tb ; ð9Þ
qa ¼ ½qb  C 2a  ðC 1a  C 2a Þ þ DP   tb ; ð10Þ

where q is the linearized aerodynamic load, qb the hot gas density between the blades, C the hot gas velocity, tb
the pitch of the moving blades cascade, DP the pressure drop, a the axial direction index, u the peripheral
direction index. All of these parameters are functions of rotating radius.
P m_
By using equations (11) and (12) and considering qðP ;T Þ ¼ 0:288T from the ideal gas equation, C ¼ qA from
continuity equation, tb from measured dimensions of cascade and DP from manufacturer documents, the axial
force was calculated about 226.13 N as the agent of thrust generator. Moreover, the peripheral component
was calculated about 1000.6 N as the agent of rotational torque. [10]
Z Tip
Fu ¼ qu dr; ð11Þ
Hub
Z Tip
Fa ¼ qa dr: ð12Þ
Hub

The resultant of aerodynamic forces was loaded on the surface of the airfoil as a distributed pressure.

4.2. Finite element (FE) modeling

The bladed-disc is a kind of cyclic-symmetric structure, which consists of 90 blades. The buckets are
inserted into the steeples on the disc via straight, axial, fir-tree shaped roots (Fig. 16). Finally the buckets will
be blocked by locking pins. However, for simplifying the analysis of FE model an assembly of a bucket on a
small portion of disc was created by using the MDT software package. The study was continued by utilizing
the ANSYS finite element code. The generated FE model consists of 4128 solid elements (solid 186 and solid
187):

10 – Node quadratic tetrahedron (solid 187) elements were used to mesh the blade.
20 – Node quadratic hexahedron (solid 186) elements were used to mesh the disc section.

In addition, to model the mechanical interface of adjacent surfaces between disc grooves and root of blade,
the high-order surface-to-surface contact element was defined.
1124 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

Fig. 16. (a) The view of a circular arrangement of assembled model consisting of a disk section and buckets. (b) A single bucket.

Due to the circular symmetry of problem, the lowest surface of the disc portion was selected to fix the
model radially in a cylindrical coordinate. Centrifugal load was simulated by defining a rotating condition
with a constant speed equal to 5100 rpm around the center axis of previously defined cylindrical support.
As aforementioned before the resultant of aerodynamic forces was loaded on the surface of the airfoil as a
distributed pressure.
By attention to the failure’s sings, the method of FE modeling, were selected. In fact, due to results of
fractographic investigation and present of striation and fatigue sings, the problem was directed to vibrat-
ing analysis for determining the mechanical reason of failure and evaluating the values of detrimental
stresses. Hence, the FE model was analyzed in two vibration methods; firstly, a modal analysis was per-
formed for studying the natural mode shapes and the related frequencies. The basis of this analysis
method is established on an undamped vibrating system that is presentable in a general form as following
equation:
½Mf€
ug þ ½Kfug ¼ 0; ð13Þ
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1125

where [M] is the matrix of mass, [K] the matrix of stiffness and {u} the displacement vector as a function of
time.
In the above equation, the term of damping force is omitted and by solving it, the eigen values and vectors
will be obtained. These parameters can only demonstrate the shape of structural deformation or relative dis-
placement and also, the stress distribution contours, relatively. However, via the solution of problems by this
method the values of natural frequencies can be calculated exactly.
In the second stage, a harmonic analysis was performed to investigate the frequency response of bucket and
lock-pin in a wide range of exciting frequencies. The real values of distributed stresses were determined along
the analysis of parts. This method is applied to calculation the steady state response of a vibrating system,
which is influenced by a sinusoidal stimulus. The general form of a damped vibrating system is defined as
the following equation:
uðtÞ g þ ½Bfu_ ðtÞ g þ ½KfuðtÞ g ¼ fP ðtÞ geixt ;
½Mf€ ð14Þ
where {P(t)}eixt is a load vector, which is frequency function of time and [B] the matrix of damping. Harmonic
analysis can result either the values of displacement, velocity and acceleration of nodes or the force and stress
of elements [11].

4.3. Results of the finite element calculations

To examine the dynamic behavior of blade, the created FE model was investigated in three methods; static,
modal and harmonic analyses. Via static analysis, the stress value cause to centrifugal force was calculated
(Fig. 17). The FEM results showed good agreement with the results obtained from analytical calculations
(Fig. 14). The stress values of leading and trailing edges, in the broken position are 186.2 MPa and
127.2 MPa, respectively. Because of the constant operating speed of turbine rotor the value of centrifugal force
and generated tensile stresses are constant too, which tend to break the blade in the most critical cross-section
area of blade airfoil.
During modal analysis some natural mode shapes and frequencies, were calculated. According to Fig. 18,
the third mode in frequency of 2094.47 Hz shows a bending type in the peripheral direction with maximum
deflection, approximately in the 0.6 length of blade. The maximum displacement has occurred in the leading
and trailing edges with a good agreement to the elevation of fractured section.

Fig. 17. Result of static analysis shows the stress values due to centrifugal load, which is reduced from, root to tip of bucket. In addition, a
figure of fractured blade is presented for comparing with the simulated FE model.
1126 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

Fig. 18. Result of modal analyses shows the deformed shape due to the third vibrational mode of bucket. The frequency mode is bending
type in the peripheral direction, 2094.47 Hz. The view of FE model comparing to the fractured specimen demonstrates a good agreement
with the breaking path at the top.

By means of harmonic analysis the distributed dynamic stresses, on the blade were correspondingly calcu-
lated for the third natural mode. Fig. 19 shows three labeled stress values, which are adjusted to trace the frac-

Fig. 19. The harmonic analysis shows the value of dynamic stresses distributed along the bucket. The labeled points were selected to show
the value of stresses corresponding to the fractured sample. The yellow arrows demonstrate precession of breaking edge.
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1127

ture path in three different elevations. The dynamic stresses have been increased, from leading toward trailing
edge.

4.4. Campbell diagram and excitation factors

During the mechanical analyses, it is found a distinct accordance between the results of vibrational analyses
and the shape of fracture across the blade airfoil (Fig. 18). In fact, it is proved that the failed bucket has been
involved to a resonant condition by the third vibration mode for a short or long time. However, the present of
striation signs on the ruptured area shows the existence of a high frequency vibrating load as a fatigue agent to
progress the crack above the mid-span of blade. Substantially resonant conditions are dangerous and blades
should be protected from those conditions. In the other hand, there are many sources, which cause to vibrate
blades in the natural modes and involving them to resonant conditions. Some of these vibrating sources are
the inherent specifications of turbo engines.
Marscher [12] has grouped the excitations factors into three categories: those involving free vibration, those
creating forced vibrations, and those involving self-excited instability phenomena. Common excitations in
each of these categories are listed below, with the order of excitation (e.g. number of cycles per second relative
to rotor speed, N, or vane passing frequency, M) noted if one is characteristic. This is important because res-
onance to sinusoidal excitation can only occur when the order of excitation times the running speed equals a
mode frequency.

Vibration types Sources and harmonics

Free 1. Intermittent stalls
2. Surges
3. Liquid slugs
Forced 1. Unbalance (1  N)
2. Vane passage (I  M)
3. Volute tongue (1  N or 2  N)
4. Plenum acoustics
5. Base excitation harmonics
6. Piping dynamics
7. Combustors (No. of combustors  N)
8. Rotor/stator rubbing
9. Rotating stall (1/3  N)
10. Coupling misalignment (1  N and 2  N)
Self-excited 1. Flutter/vortex shedding
2. Loose hub connection
3. Shaft hysteresis (1/2  N)
4. Oil film whip (1/2  N)

The base excitation harmonics due to nozzle turbulences or aeroelastic disturbances can relatively influence
in the high-order of per revolution harmonics.
In the other hand by means of a Campbell diagram it was recognized that the 24th and 25th per revolution
harmonics are the nearest to the working speed of rotor for excitation the third natural mode of blade
(Fig. 20).
During the log sheet surveys, (the frequency of electric output power) some abnormal operating conditions
were found. The machine, in the rotating speed of 3000 rpm, should generate the standard frequency value of
50 Hz. The generator rotor is coupled to a reducing gearbox, which is driven by the gas turbine rotor. By
means of a reduction factor (i = 3000/5100) for gearbox, the speed of turbine can easily be traced from the
output frequency in the log sheets. The results of log sheet surveys are presented in Fig. 21.
1128 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129

H25
2200 H24
Mode III
H23
2000 H22
Mode II
Frequency [H z]
H21
1800 H20
H19
Mode I
1600

1400

Working
Speed
1200

1000
50 55 60 65 70 75 80 85 90
Rotor Speed [rps]

Fig. 20. Campbell diagram for the second stage of turbine blades.

50.5
2004/08/04

2004/08/05
Electric Output Frequency [Hz]

2004/08/06

50 2004/08/07

2004/08/08

2004/08/09

2004/08/10
49.5 2004/08/11

2004/08/12

2004/08/13

2004/08/14
49
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Monthly Dates
Daily Hours

Fig. 21. A daily record of turbo generator electric frequency output for ten days before the failure event.

As it is seen, the machine has been utilized with low derivations of output frequency except in the two time
limits in which the frequency is decreased to 49.4 Hz (83.98 rps). The mentioned problems have been occurred
two days before the failure (14 August 2004). However, the first problem was occurred for a short time interval
at 9:00 PM on 12 August, but the second problem has been occurred for 1-h duration from 3:00 PM until 4:00
PM on 13 August, exactly one day before the failure. In the other hand the related rps of these two malfunc-
tions, comparing to the Campbell diagram is very close to the dangerous limit of twenty fifth per revolution
harmony for excitation the third vibrational mode of blades.

5. Conclusion

The catastrophic failure of the blade has occurred by the following sequence:

1. Formation of the non-protective nickel and cobalt oxides.

2. Formation of the chromium sulfide and depletion of the alloying elements.
 E. Poursaeidi et al. / Engineering Failure Analysis 15 (2008) 1111–1129 1129

3. Degradation of the metal beneath the scale.

4. Progression of the pitting over the concave and convex surfaces of the blade.
5. Deepening of the pits at the leading and trailing edges due to the bending stresses.
6. Development of the interdendritic corrosion at the leading edge and initiation of the crack.
7. The propagation of crack by the fatigue mechanism because of the vibration of blade in a resonant
condition.
8. Reduction of the cross-section area and the final fracture at the trailing edge.

Acknowledgments

The authors would like to thank Mr. Kaluti and Mr. H. Mohammadi for their financial support through
investigation of the projects. In addition, Mr. Ahmadi, Mr. Razavi and Mr. Nemati for their contributions to
gathering the mechanical and metallurgical data and Mr. Oghbaee for his contribution to preparing the failure
photos. Also Mr Kamali and Mr. J. Poursaeidi for contributing in type setting and metallurgical test prepa-
ration respectively.

References

[1] Brooks CR. Heat treatment, structure and properties of nonferrous alloys. New York: ASM International; 1984.
[2] Sims CH, Stoloff N, Hagel W. Superalloys II: High temperature materials for aerospace and industrial power. New York: Wiley-
Interscience Publication; 1987 [p. 46–52].
[3] Mazur Z, Luna-Ramirez A, Juarez-Islas JA, Campos-Amezcua A. Failure analysis of a gas turbine blade made of Inconel 738LC
alloy. Eng Fail Anal 2005;12:474–86.
[4] Carter TJ. Common failure in gas turbine blades. Eng Fail Anal 2005;12:237–47.
[5] Barbosa C, Nascimento JL, Cominha IMV, Abud IC. Microstructural aspects of the failure analysis of nickel base superalloys
components. Eng Fail Anal 2005;12:348–61.
[6] Viswananthan R. An investigation of blade failures in combustion turbines. Eng Fail Anal 2001;8:493–511.
[7] Eliaz N, Shemesh G, Latanision RM. Hot corrosion in gas turbine components. Eng Fail Anal 2002;9:31–43.
[8] Hou J, Wicks BJ, Antoniou RA. An investigation of fatigue failure of turbine blade in a gas turbine engine by mechanical analysis.
Eng Fail Anal 2002;9:201–11.
[9] Chang JC, Yun YH, Choi C, Kim JC. Failure analysis of gas turbine buckets. Eng Fail Anal 2003;10:559–67.
[10] Frolov Kostyuk V. Steam and gas turbines. Moscow: Mir Publishers; 1988 [in Russian].
[11] MSC-PATRAN software help, translated by Talebi H, Farzadi H. Mehri A. Tehran, Iran: Taraah publication co.
[12] Sawyer’s gas turbine engineering handbook, vol. 1, 3rd ed. Turbomachinery International Publication; 1985 [chapter 7].