Influence of pitting on the fatigue life of a turbine blade steel

S. Z H O U a n d A . T U R N B U L L
National Physical Laboratory, Teddington, Middlesex TW11 OLW, UK

Received in final form 27 May 1999

A B S T R A C T The role of pits as stress raisers and their influence on fatigue life has been investigated
for a 12Cr turbine blade material. A particular feature of this work was the establishment
of an electrochemical procedure for generating pits with ‘controlled’ pit depth and
low density.
Pits grown under laboratory conditions were partially spherical in shape and simulated,
in general appearance, those observed in service. In terms of the threshold stress
intensity factor, the results supported the concept of pits acting as effective cracks of
the same depth, provided that a short crack model based on an effective crack length
is used.
Keywords corrosion-fatigue; pitting; short cracks; 12Cr stainless steel.

when the pits reach a critical size so that the stress

INTRODUCTION
intensity factor range exceeds the threshold for crack
Over the past 30 years or so there has been a number growth and the growth rate of cracks exceeds the growth
of examples of failures of steam turbines as a consequence rate of pits. A major concern is that the threshold stress
of the development of cracks from corrosion pits in both intensity for stress corrosion cracking or threshold stress
blade and disc materials.1–3 The consequences are often intensity range for corrosion fatigue determined from
catastrophic as failure of a single blade destroys long cracks of fracture mechanics specimens cannot be
additional rows of vanes and blades.4 Numerous labora- applied to short cracks initiated from pits. Moreover,
tory studies have been conducted to address this accelerated testing is usually necessary and in corrosion
problem.5–12 fatigue testing this is usually achieved by increasing the
Pits can promote crack development, in general, by frequency. In an aqueous environment, this may lead to
acting as stress concentrators, by creating a local solution crack initiation on a time scale which is short compared
chemistry conducive to cracking and by removing to that for the development of pits and correspondingly
impediments to hydrogen ingress through dissolution of may underestimate the impact of pitting. For this reason,
oxide films. Pit nucleation and development need not there is considerable virtue in pre-pitting fatigue test
occur under ‘normal’ operating conditions but may be samples in a controlled manner to the relevance depths.
induced during an excursion, e.g. an outage, a transient This has the additional advantage of identifying critical
in temperature or water chemistry, or under intermittent pit sizes for crack nucleation.
wetting conditions. In low-pressure turbines, pitting Various methods of pre-pitting have been investigated.
develops during exposure to aerated condensate con- Pits have been generated mechanically,17 which has the
ditions typical of that found during start-up and shut- advantage of controlling the ‘pit size’ and density.
down periods. There is no evidence of pits propagating However, the pit shape and the local microstructural
in on-load environments.7 features may not be the same as for pits generated
Lifetime prediction is complex because of the need to naturally in solution. A slightly more relevant method
account for pit initiation, pit growth, crack initiation, of generating pits, using electrochemical means, has
short crack growth and long crack growth.13 Several been developed at the All-Union Heat Engineering
models have been proposed for life prediction of cracking Institute.18 The working surface of a smooth specimen
initiated from pits, based on the kinetics of pit growth was coated with lacquer and paraffin wax in which
and crack growth.14–16 The condition for the transition openings of a given size were left in a specific pattern.
of cracks from pits is generally defined phenomenologi- The specimen was then immersed in a solution of 10%
cally using a LEFM approach, i.e. the transition occurs FeCl3 (pH 2.2). An unspecified current was applied

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093 1083
1084 S. Z H O U a nd A . T U R N B U L L

Table 1 Composition of Jethete M152 steel

Cast C Si Mn P S Cr Mo Ni N V Fe (per cent by weight)

B9432 0.11 0.18 0.65 0.015 0.004 11.50 1.56 2.52 0.028 0.29 bal
C4352 0.11 0.23 0.81 0.013 0.002 11.58 1.67 2.67 0.032 0.30 bal

continuously, causing anodic dissolution of the metal in generation procedure and the impact of pitting on fatigue
the exposed region. This technique can be used to limits, which was addressed in the present study, would
generate pits with a specific density and geometry. not be affected by the difference in heat treatment.
However, the environmental test conditions are very
severe. In addition, the sites at which pits initiate are
Electrochemical and pitting tests
determined by the location of the openings in the
coating, rather than by surface defects, e.g. inclusions, The initial electrochemical and pitting tests were con-
as are the pit diameter and geometry. ducted on a disc of the Cast B9432 steel in order to
Pits generated by pre-exposure have been used in establish the procedure prior to pitting of the cylindrical
some studies. Pre-generated pits can be produced by fatigue specimens. The working area of the disc electrode
exposing the specimen to simulated environments.6,19,20 was 0.9 cm2 .
The depth of pits required was achieved by using an The fatigue specimens were made from Cast C4352
appropriate exposure time. However, no attempt was in accordance with the ASTM standard21 and had an
made to optimize the exposure procedure so that a overall length of 200 mm and a shoulder diameter
controlled depth and density could be obtained in a of 12.6 mm. The gauge length was 20 mm and the
repeatable manner. diameter 5 mm.
The objectives of the present study are: to establish an For both disc and cylindrical specimens, the surface
electrochemical procedure for growing pits in a turbine was dry-ground to a 2400 SiC grit finish, and rinsed in
blade steel to a controlled depth, with a shape similar to distilled water, ethanol and acetone before use. For the
that in service; to assess the impact of pit depth on fatigue
life; to assess the validity of the fracture mechanics approach
in characterizing the impact of pit depth.

EXPERIMENTAL

Material composition and mechanical properties

The material used for this work was a gas turbine
martensitic steel, Jethete 152, supplied by British Steel,
in the form of a bar 1 m in length, 10 cm in diameter.
Two batches of material (Cast B4932 and C4352) were
used. The compositions of the steels are listed in Table 1.
The steel was ingot cast, rolled, annealed and turned.
The microstructure of the material is shown in Fig. 1.
The steel was martensitic with a mean prior-austenite
grain size of ~25 mm. The mechanical properties of the
Fig. 1 Photograph showing the microstructure of Jethete M152
material are listed in Table 2. The data are the average steel (×500).
of three measurements in each case. The differences in
the mechanical properties of the two batches of material Table 2 Mechanical properties of Jethete M152 steel
were relatively minor. It should also be noted that the
steel was tested in the annealed condition as received. 0.2%
In actual service, in the form of machined or forged Proof
turbine blades, the steel would be hardened and tem- stress UTS Elongation Reduction Hardness
pered to a slightly higher strength level. For example, a Cast (MPa) (MPa) (%) in area (%) (Rockwell C)
typical turbine manufacture’s specification for Jethete
B9432 728 903 22.3 65.3 28
M 152 is 780 MPa for minimum 0.2% proof stress and C4352 750 937 22.1 65.1 30
930 MPa for minimum UTS. However, the issues of pit

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
 I N F L U E N C E O F P I T T I N G O N FAT I G U E L I F E O F T U R B I N E B L A D E S T E E L 1085

fatigue specimens, it was critical to ensure a uniform

RESULTS
state of surface finish over the whole exposed surface to
control the sites for pit initiation. The fatigue specimens
Electrochemical measurements
were ground in the transverse direction.
The solution used was 0.1 M NaCl, prepared from AR Electrochemical polarization measurements (Fig. 2) were
grade NaCl and distilled water. The solution was open conducted at 60 °C (scan rate: 1 mV/s) for two values of
to the air. The temperature was controlled at 60±1 °C. surface roughness, 2400 and 1200 grit finish. Three tests
The initial pH of the solution was 5.5±0.1 and the final were conducted for each surface finish. The key features
value <6.4±0.1. A platinum counter electrode was used are the increase in pitting potential as the surface
and the potential measured with respect to an external becomes smoother and the lack of dependence of the
saturated calomel electrode (SCE) connected to the test repassivation potential on the surface finish. This is to
cell through a salt bridge. All potentials quoted are with be expected as the rougher specimens will have more
respect to SCE at 20 °C. sites for localized chemistry changes conducive to passiv-
Crevice corrosion of the disc and cylindrical specimens ity breakdown. At the test temperature of 60 °C, the
was prevented by using a 0.25-L cell with an Avesta- pitting potential (the potential at which the anodic
type seal.22 Distilled water is pumped (2 mL/h) into the current density reaches 10 mA/cm2 during the forward
region of the seal and this limits the build up of an scan) was ~0.03 V (SCE) for the specimen with the
aggressive environment. A solution of increased strength 2400 grit finish. This potential is not very far above the
was added continuously to the bulk solution to compen- corrosion potential of the material, −0.09 V (SCE)
sate for the dilution, albeit small, due to the ingress of (Table 3), when exposed at 60 °C over a period of 60 h
distilled water in the seal region. (Fig. 3). No pitting was observed in the latter test.
To limit the spread of pitting over the fatigue speci-
men, the Luggin capillary for the reference electrode
and the platinum counter electrode were located close
to the centre of the specimen.

Fatigue tests
The fatigue tests were performed under uniaxial tension–
tension loading in air at ambient temperature (25±2 °C)
using a servo-hydraulic fatigue machine in accordance
with BS 3518.23 The relative humidity of the air was
below 50% for all tests. The fatigue frequency used was
primarily 25 Hz although a few tests were conducted at
1 Hz. The stress ratio was 0.1. The specimens without
pre-pitting were ground transversely to a final finish of
400 or 2400 grit. The surface profile24 was measured
using a laser profilometer. Measurements were made at
seven locations on the specimens with a 400 grit finish Fig. 2 Effect of surface finish on the pitting and repassivation
potentials of Jethete M152 steel in 0.1 M NaCl solution at 60 °C.
and at four locations on the specimens with a 2400 grit
finish. The length of profile for each individual measure-
ment was 2 mm. The arithmetical mean deviation of the
Table 3 Corrosion potential, pitting potential and repassivation
assessed profile, Ra , was 0.87±0.03 mm for the 400 grit
potential of Jethete M152 steel in 0.1 M NaCl solution at 60 °C
dry-ground specimens and 0.25±0.01 mm for the 2400
grit specimens. The corresponding maximum height of Ecorr Epitting Erepassivation
the profile, Rz , was 10.37±1.97 mm and 2.63±0.28 mm Surface (V/SCE) (V/SCE) (V/SCE)
for the 400 and 2400 grit finish, respectively.
The depth of the pit and the width of the pit on the 2400 grit
surface were measured using a travelling microscope finish −0.09 0.03 −0.13
1200 grit
(resolution 1 mm) prior to fatigue testing and also on the
finish — −0.01 −0.12
fractured surface after testing. The difference in pit
depths between the two measurements was <5%, but Note that the corrosion potential was measured after 60 h exposure;
the depth measured on the fractured surface is intrin- the pitting potential and repassivation potential were measured
sically more relevant and is quoted in all fatigue results. using a potentiodynamic scan at 1 mV/s scan rate.

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
1086 S. Z H O U a nd A . T U R N B U L L

Fig. 3 Variation of the corrosion potential of Jethete M152 steel in Fig. 4 Time dependence of the maximum pit depth of Jethete
0.1 M NaCl solution at 60 °C. M152 steel in 0.1 M NaCl solution at 0.0 V (SCE) and 60 °C. [The
initial pit depth of 35 mm at t=0 was generated by stepping the
potential to 0.1 V (SCE) for 30 s.]
Establishment of pit generation procedure
In order to initiate pits, the electrode potential must be The surface appearance of the pits generated was
raised above the pitting potential. Maintaining the poten- approximately circular, as shown in Fig. 5. The pit
tial at this value to achieve the required pit depth would dimensions in two directions as a function of depth
simply lead to generation of a relatively high density of were measured using the ‘layer removal’ technique and
pits of variable size. This is not desirable as interaction the results are shown in Fig. 6. The pits are partially
of pits may cause difficulty in defining the relationship spherical and not dissimilar to those observed in
between the fatigue limit and the pit size. Also, in service.4,9
establishing the effect of pit depth on fatigue life, more The pit generation procedure described above was
extensive testing would be required because there is no applied to the cylindrical fatigue specimens in order to
control over the initial pit size. However, if the applied generate three discrete maximum pit depths. After 30 s
potential is stepped, after pit initiation, to a value below at 0.1 V (SCE), the depth of the deepest pit was ~35 mm;
the pitting potential but above the repassivation potential after stepping to 0.0 V (SCE) and holding at this poten-
for the deeper pits, no fresh pits will be generated and tial for 5 min the maximum pit depth was 110 mm;
only pits greater than a certain depth will continue to 30 min at 0.0 V (SCE) gave pits with a maximum depth
grow. Maintaining a constant current above the pitting of ~250 mm.
potential would probably achieve the same aim.
After preliminary tests, the experimental procedure to
generate pits was established as follows.
The specimen was immersed at open circuit for
1 h, at which time the open circuit potential was
~−0.15 V (SCE).
A potential of 0.1 V (SCE) was applied for 30 s to
initiate pits.
The potential was stepped down to 0 V (SCE) to
allow the pits to grow. The time at 0 V (SCE) was
adjusted depending on the pit depth required.
The dependence of the maximum pit depth on the
elapsed time at 0 V (SCE) is shown in Fig. 4. From
linear regression analysis, the pit growth can be expressed
by
a=btb (1)
where a is the maximum pit depth in mm, t is the time Fig. 5 Photograph showing the pit generated on Jethete M152
in minutes, and b and b are 48.03 and 0.53, respectively. steel in 0.1 M NaCl solution at 0.0 V (SCE) and 60 °C.

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
 I N F L U E N C E O F P I T T I N G O N FAT I G U E L I F E O F T U R B I N E B L A D E S T E E L 1087

Fatigue tests
Fatigue tests were performed on specimens with and
without pre-pitting to establish a relationship between
the fatigue limit and pit size. The fatigue limit is defined
as the value of the stress range below which the fatigue
Pit depth (µm)

specimen would be expected to endure an infinite

number of stress cycles with a stated probability.21 In
practice, it is often defined as the value of the stress
range below which the specimen does not fail above a
certain number of cycles. In the present work, this
number of cycles has been defined as 107.
Fatigue tests in air were conducted on Jethete M152
steel (Cast B9432) specimens with a 400 grit dry-ground
ground finish at two frequencies of 1 and 25 Hz
[Fig. 7(a)]. There was no significant effect of fatigue
Pit width (µm) frequency on the fatigue life in air for the stress range
tested, as expected. Hence, all subsequent tests were
conducted at 25 Hz.
The fatigue limit of the two batches of materials was
similar, as shown in Fig. 7(b). The fatigue limit was
markedly increased when the specimen was ground to a
2400 grit finish. Indeed, with a stress ratio of 0.1, the
maximum stress at the fatigue limit was above the 0.2%
proof stress.
Pit depth (µm)

Fatigue tests were conducted on specimens pre-pitted

to maximum pit depths of 35, 110 and 250 mm. The
variability in the depth of the pits initiating cracking was
±5%. The results are shown in Fig. 8. All of the tests
on pre-pitted specimens were carried out using the
second batch of material (C4352).
The estimated fatigue limits of the specimens with
and without pits are summarized in Table 4. The uncer-
tainty in the fatigue limit is the difference between the
maximum stress at which the specimen did not fail and
the minimum stress at which the specimen failed. The
Pit width (µm) significant impact of pit depth on the fatigue limit is
apparent. However, the surprising feature of the results
is the similarity in fatigue limit for specimens with a 400
Fig. 6 Pit profile measured using the layer removal method in grit dry-ground finish and those pre-pitted to a depth
(a) the X direction and (b) the Y direction. of 35 mm. The Ra value for the 400 grit finish was
0.87±0.03 mm, while the Rz value was 10.37±1.97 mm.
Fractographic examination revealed that there were
grinding defects on the fracture surface of specimens
Due to the intrinsic random nature of pit initiation, with the 400 grit dry-ground finish. The depth of the
it is difficult to control the distribution of pit density defects was measured on four specimens, which had
although the positioning of the Luggin capillary and fractured just near the fatigue limit, and was in the range
counter electrode close to the centre of the specimen 28–38 mm. The defect on the specimens fractured at a
had the effect of restraining the deepest pits to the gauge higher stress range was more difficult to detect and
length as a result of the potential drop of the solution. measure due to the deformation at the edge of the
The pit density was reasonably repeatable, usually fractured surface. Variation in the depth of defect on the
one–two pits of significant depth on a specimen of area 400 grit dry-ground specimens is the probable expla-
0.9 cm2 , with the distance between the pits at least five nation for the scatter of the cycles to failure data for the
times the pit diameter. B9432 material (Fig. 7).

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
1088 S. Z H O U a nd A . T U R N B U L L

Stress range, Ds (MPa)

Fig. 8 Effect of pitting on the fatigue life of Jethete M152 steel

(R=0.1, f =25 Hz).

Table 4 The fatigue limit of Jethete M152 steel with and without
Cycles to failure, Nf pits (stress ratio: 0.1; frequency: 25 Hz)

Batch Surface Fatigue limit (MPa)

Cast B4352 400 grit dry ground finish 573±13

2400 grit ground 709±11
Cast C4352 400 grit dry ground finish 574±11
2400 grit ground 709±11
pit of 35 mm depth 619±11
Stress range, Ds (MPa)

pit of 110 mm depth 495

pit of 250 mm depth 371±11

Table 5 The surface residual stress of the specimens after 400 and
2400 dry grinding, assuming no shear stress is present

Condition Direction Residual stress (MPa)

400 grit Long −365±12

Hoop* −468±17
Long −205±16
2400 grit Hoop −239±11

*The average of two measurements.

Cycles to failure, Nf
Fig. 7 S–N data of Jethete M152 steel in air: (a) effect of the DERA and the results are listed in Table 5. The X-ray
loading frequency; (b) effect of surface finish. penetration depth is ~10 mm; hence, these values rep-
resent the average in the first 10 mm of the steel. As can
The concern is that the surface profiling proved less be seen from Table 5, there is a significant compressive
effective than fatigue testing in probing for defects. It is residual stress in both longitudinal and hoop directions,
possible that the spatial resolution (the distance between the magnitude being greater for the 400 grit finish. The
data points) of the laser profilometer, 2 mm, is not effect of the compressive residual stresses would be to
sufficiently high to accurately measure the depth of the lower the mean stress (see later).
deepest grinding defect, which may be partly closed due
to the compressive surface residual stress induced by
DISCUSSION
grinding.
Residual stress induced by the surface finishing process Providing a general framework to predict the impact of
is always a concern in fatigue testing. For this study, the corrosion pits on the fatigue behaviour of the turbine
residual stress was measured using X-ray diffraction at blade steel is non-trivial. The pit shape and topography

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
 I N F L U E N C E O F P I T T I N G O N FAT I G U E L I F E O F T U R B I N E B L A D E S T E E L 1089

can change with pit depth; analytical solutions for the

impact on stress intensity factors of notch-like defects,
e.g. pits, are based on two-dimensional geometry only;
the pit depths are small, local plasticity at the notch root
may prevail and thus linear elastic fracture mechanics
(LEFM) may become inapplicable.
Treatment of the impact of notches on the fatigue
limit is based usually on two limiting situations.25 As the
notch root radius becomes large, the specimen tends to
become plain; as the root tends to zero, the notch may
be equated to a crack of length equal to the depth of
the notch. For relatively mild, blunt notches, once the
crack has initiated, the bulk stress level is sufficiently
high for the crack to propagate to failure. For sharper
notches, the fatigue limit is based not on initiation but
on the higher level of stress required for propagation.
Non-propagating cracks are then observed.
A fracture mechanics approach has been applied to
both cases. The assumption is that there are inherent
defects of some small size on both plain and notched
specimens. The issue is whether cracks will develop from
the defect and, if developed, will they continue to grow.

Pits as blunt notches

The stress concentration factor for a notch, assuming
elasticity at the notch root, is defined by

snotch
Kt = (2)
snom

where snom and snotch are the normalized stress and the
stress at the root of the notch, respectively.
For a hemispherical pit, finite element analysis indi-
cates a stress concentration factor of 1.8 independent of
pit depth, assuming constancy of geometry and a large
radius of the cylindrical specimen. The pits in this study
are only a crude approximation to this shape but a stress
concentration factor of about this value would be reason-
able. From Table 2, this would imply local yielding for
all but the deepest pit. Hence, treatments based simply
on an elastic stress concentration factor would be
inappropriate.
Fig. 9 SEM photographs of the fracture surface of fatigue cracks
Smith and Miller25 account for plasticity at a notch initiated from pits of depth: (a) 35 mm; (b) 110 mm; and (c) 250 mm.
root and proposed a relationship for the fatigue strength
reduction factor (fatigue limit for a notch specimen
divided by fatigue limit of a plain specimen) which was 110 and 250 mm. Because there was some narrowing of
subsequently updated by Cameron and Smith26 to give the pit diameter near the mouth, the maximum diameter
was selected in defining 2c. The fatigue strength

A SB a 0.5 reduction factor, Kf , calculated from Eq. (3), assuming

Kf = 1+4.67 (3) the pit root radius is half of the pit diameter, is shown
c
as a function of pit depth in Fig. 10. It is apparent that
where a is the notch depth and c is the radius for a pit. this relationship does not correlate well with the exper-
Figure 9 shows microphotographs of the fracture sur- imental data. Although the quantitative basis of Eq. (3)
faces of fatigue cracks initiated from pits of depth 35, is for a through-thickness notch, it might be expected

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
1090 S. Z H O U a nd A . T U R N B U L L

overall ‘crack’ length including the pit is small; a non-

propagating crack formed at the pit may have an overall
length which includes the pit but does not have the
plastic wake history of a crack of the same length, the
crack will be contained within the plastic field of the pit.33
In order to account for the breakdown of the LEFM
concept for small crack sizes, El Haddad and co-
workers34–36 proposed an empirical relationship for the
stress intensity factor range based on the notion of an
effective crack length (a+a0 ):
DK=aDs 앀p(a+a0 ) (6)
and,
DKth =aDsth 앀p(a+a0 ) (7)
Fig. 10 The impact of pits on fatigue strength reduction factors. where DK is the range of the stress intensity factor, Ds
The number in brackets is the value of a/c for the particular pit
is the stress range and a0 is a constant.
depth.
This relationship has been shown to describe short
crack and pit behaviour well for a range of systems, and
that the broad trend should be predicted. Other relation-
provides a link to long cracks.31,34–37 The a0 value can
ships27,28 proposed to account for blunt notches based
be evaluated from the limiting condition where the
on root radius also cannot be reconciled with the data
physical crack size, a, approaches zero, i.e. a0 &a and
of Fig. 10.
Dsth =Ds0 , giving

A B
2
1 DKth
Pits as effective cracks a0 = (8)
p aDs0
The supposition that pits behave like cracks in determin-
ing threshold conditions for cracking is not new or where Ds0 is the fatigue limit of a smooth specimen.
unusual in the context of a fatigue limit based on The parameter, a0 , is essentially a fitting parameter in
propagation. Eqs (6) and (7), and its physical meaning is not well
In that context, the range of the stress intensity factor understood.33 Because DKth and Ds0 are dependent on
has been used16,17,29 to account for the role of pitting in material characteristics, a0 is considered to be a material-
defining threshold using linear elastic fracture mechanics related parameter. However, a0 is appreciably larger
(LEFM), viz. than any microstructural characteristic and cannot be
related to the characteristic length of the materials’
DKth =aDsth 앀pa (4) microstructure.37
where DKth is the range of the threshold stress intensity Equation (7) can be expressed as:
factor, Dsth is the threshold fatigue strength, a is the pit 2
DK th
depth and a is a geometry factor. For semicircular surface a= −a0 (9)
a2 pDs2th
cracks, a is approximated by 0.67.30–32
Lindley et al.6 considered a pit as a semielliptical From Eq. (9), it can be seen that DKth and a0 can be
defect and derived the following relationship: determined readily by applying linear regression analysis
Dsth 앀pa[1.13−0.07(a/c)1/2 ] to a plot of a versus Ds−2th .
DKth = (5) Assuming again a value for a of 0.67, a value of 82 mm
[1+1.47(a/c)1.64 ]1/2
for a0 was obtained from the data listed in Table 4. It
where 2c is the surface length of the defect. Equation should be emphasized that the values of both a0 and
(4) is a particular case of Eq. (5), i.e. a/c=1. DKth are dependent on the applicability of the parameter
The average aspect ratio of pits from all tested speci- a and its constancy with pit depth. In the same context,
mens and the threshold stress intensity factor calculated the cyclic plastic strain existing at the base of the pit
from Eqs (4) and (5) are listed in Table 5. may require a more rigorous treatment than the simple
Because Eqs (4) and (5) are based on linear elastic elastic analysis of Eq. (6).34,35
fracture mechanics, DKth should be a constant indepen- The value of the threshold stress intensity factor range
dent of crack size. This is clearly not the case. calculated in this way is included in Table 6. A threshold
The limitation of LEFM is not surprising as the independent of pit size is obtained, consistent with

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
 I N F L U E N C E O F P I T T I N G O N FAT I G U E L I F E O F T U R B I N E B L A D E S T E E L 1091

Table 6 Aspect ratio of pits and the calculated threshold stress Recalling that the residual stress induced by grinding
intensity factor range to a 400 grit finish was greater than that for the 2400
grit finish, the similarity of fatigue limits for the 400 grit
Aspect ratio Fatigue DKth (MPa앀m)
Pit depth (a/2c) limit
finish with machining defects (28–30 mm) and the 2400
(mm) (average) (MPa) Eq. (4) Eq. (5) Eq. (7) grit finish with 35 mm pits would tend to suggest that
for this depth of defect residual stress was not significant
35 0.5 619 4.35 4.35 7.94 enough to influence the fatigue life.
110 0.72 495 6.16 5.06 8.14 Lebedeva et al.18 also studied the effect of pits on the
250 0.73 371 6.97 5.61 8.02 fatigue limit of turbine blade materials (12–13Cr steel)
but using a different pre-pitting procedure as outlined
above. The aspect ratio of pit depth to diameter was
Eq. (7). As the pit depth increases, the threshold stress ~0.3 and the pit diameters were 0.4, 0.5 and 1.8 mm.
intensity factor calculated from Eq. (3) will approach The data of Levedeva et al. have also been used to test
that evaluated from Eq. (7). The contribution of a0 is the El Haddad approach and are included in Fig. 11.
less than 10% when the pit depth reaches 400 mm and Because the data were limited, the fit was obtained using
thus such a pit can be treated as a long sharp crack. the value as a0 as one of the parameters. Again, Eq. (7)
The applicability of Eq. (7) to the experimental data is consistent with the experimental data, but with a0 =
is also illustrated in the Kitagawa type plot38 of Fig. 11. 169 mm and DKth = 6.4 MPa 앀m. These values are sig-
When the pit depth increases, the curve calculated from nificantly different from those determined in the present
Eq. (8) approaches that from Eq. (4), i.e. the deeper pits work, viz. a0 =82 mm and DKth = 8.0 MPa앀m. The steel
(>1 mm) can be treated as long cracks using conven- tested is different, with the values of s0.2 and sUTS (640
tional linear elastic fracture mechanics. Correspondingly and 792 MPa, respectively) considerably lower than
the fatigue limit decreases progressively. those of the Jethete M152 steel (Table 2). In addition,
It might be expected that the compressive residual the assumption that a is identical may not be valid, as
stresses at the surface induced by grinding would affect the aspect ratio of the pits generated is different.
the fatigue limit of the smooth specimen, but would The concern is whether the differences are associated
have a reduced influence as the pit depth increases, as in any way with the pit generation procedure. It is
the residual stresses due to grinding are perceived to be important to establish a pit generation procedure which,
local to the surface.39,40 The solid curve in Fig. 11, fitted whilst accelerated, is not too far removed from service
using the fatigue limits of the specimens with the pits of conditions. Our approach has the merit of avoiding
35, 110 and 250 mm in depth, approaches a value environments, temperatures and applied potentials which
(750 MPa) which is greater than the fatigue limit are not too extreme with respect to service conditions.
(709±11 MPa) of the ‘smooth’ specimens with the 2400 The detailed shape of pits at the microscopic level is not
grit finish. If compressive residual stresses were to have necessarily smooth and will depend on the method of
an impact, the apparent fatigue limit measured for the generation. For instance, the aspect ratio of pits produced
2400 grit finish might have been expected to be the by Lebedeva et al.18 is different from that generated by
higher. However, this neglects the possible role of the the procedure developed in the present work (~0.3 in
surface roughness in the latter case. the former and 0.5–0.7 in the latter). However, it may
be the case that for fatigue limit determination, for
which the pit acts like an effective crack, the exact
method is unimportant unless there is a specific impact
of the detailed pit shape on the threshold. This can be
resolved by exploring other methods of generating a pit
on the Jethete steel used in this study and will be
considered for future investigation.

IMPLICATIONS
Ideally, the data generated could be used to define critical
pit sizes from the viewpoint of inspection. In practice,
non-destructive evaluation (NDE) will not usually detect
defects/cracks below 1 mm in steam turbines. It is neces-
sary then to establish a methodology for predicting the
Fig. 11 Effect of pit depth on the fatigue limit. likelihood of a crack at the NDE limit after a certain

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
1092 S. Z H O U a nd A . T U R N B U L L

operating time. Upon such detection, the requirement

REFERENCES
is crack growth rate kinetics to establish periodicity of
subsequent inspection intervals and the remnant life. 1 P. McIntyre (1989) Crack initiation at corrosion pits, review of
The first step is to establish the probability of power industry experience. In: Corrosion pitting, its practical
obtaining a pit of a size sufficient to result in crack applications: crack initiation at corrosion pits, The Institute of
Metals, London, 14 March. Br. Corr. J. 24, 103–104.
development under the loading conditions. This entails
2 R. I. Jaffee (Ed.) (1981) Corrosion fatigue of steam turbine
experimental exposure studies of pit development with blade materials. Workshop Proceedings, Palo Alto, California,
time under off-load environmental conditions. The pri- 21–24 September, Pergamon Press, London.
mary issue then is to define the latter and the impact on 3 M. O. Speidel and R. B. Scarlin (1991) In: Stress Corrosion
pit development of transient variations in exposure and Cracking and Corrosion Fatigue of Steam Turbine Rotor and Blade
temperature (i.e. off-load–on-load sequence). Materials (Edited by J. B. Marriott), Commission of the
European Communities, ECSC-EEC-EAEC and Brussels,
The transition from a pit to a crack depends on the
pp. 343–368.
local mechanical driving force, local material properties 4 S. Charkravary, P. Li, V. Tirabasso and P. C. Patnaik (1993)
and local environment. The fatigue tests in the present Improvement of fatigue and corrosion resistance of compressor
work were conducted in air and, hence, only the mechan- rotor blades of an industrial gas turbine engine. In: Proceedings
ical effect of pits on fatigue has been evaluated. A of ASM 1993 Materials Congress Materials Week ’93, Pittsburgh,
methodology for life prediction to account for pit devel- Pennsylvania, 17–21 October, USA, pp. 135–143.
opment, the transition from pitting to cracking and the 5 P. A. Coulon and H. Saisse (1993) Fatigue cracking initiation
study for blade materials, materials for blades, 12% chromium
growth of cracks under simulated service conditions is
steel (Afnor Z 19 CD 12-1). In: Workshop Proceedings, Palo Alto,
currently being developed. California, 21–24 September, Pergamon Press, London,
pp. 4-119–4-149.
6 T. C. Lindley, P. McIntyre and P. J. Trant (1982) Fatigue-crack
CONCLUSIONS
initiation at corrosion pits. Metals Technology 9, 135–142.
A pit generation procedure has been established for pre- 7 G. M. Speakes, P. Mulvihill and P. McIntyre (1987) The
implications of corrosion pitting for the refurbishment of low-
pitting fatigue specimens to a controlled depth with
pressure turbine rotor shafts. In: Proceedings of the Institute of
limited density. The key feature is to step the potential Mechanical Engineers, Conference on Refurbishment and Life
above the pitting potential to initiate pits but then to Extension of Steam Plant, London, 14–15 October, pp. 315–322.
lower the potential to a value below the pitting potential 8 R. Ebara, T. Kai, M. Mihara, H. Kino, K. Katayama and
but above the repassivation potential. If the latter poten- K. Shiota (1993) Corrosion fatigue behaviour of 13 Cr stainless
tial is chosen carefully, this will stop the growth of all steel for turbine moving blades. In: Workshop Proceedings, Palo
but the deepest pits. The elapsed time at the latter Alto, California, 21–24 September, Pergamon Press, London,
pp. 4-150–4-167.
potential is varied to control the depth of pit required.
9 R. Ebara, T. Yamada and H. Kawana (1990) Corrosion fatigue
Using this technique, pits can be grown to a specific process of 12 Cr stainless steel. ISIJ International 30, 535–539.
maximum depth with a repeatability of ±5% variation 10 M. Shalaby, J. A. Begley and D. D. Macdonald (1996)
in depth. Phenomenological aspects of fatigue crack initiation, propa-
Fatigue tests in air on a 12Cr turbine blade steel using gation in type 403 stainless steel in simulated steam cycle
pre-pitted specimens showed a reduction in fatigue environments. Corrosion 52, 262–274.
11 G. Gabetta and L. Torri (1992) Crack nucleation and propa-
strength with increasing pit depth. The fatigue strength
gation in blade steel material. Fatigue Fract. Engng Mater. Struct.
for the specimens with 35, 110 and 250 mm deep pits 15, 1101–1111.
was 87, 70 and 52%, respectively, of the fatigue strength 12 Y. Konda (1989) Prediction of fatigue crack initiation life based
of the ‘smooth’ specimen (2400 grit finish). on pit growth. Corrosion 45, 7–11.
The dependence of the fatigue limit on pit depth was 13 A. Turnbull, W. Dietzel and A. J. Griffiths (1996)
consistent with pits acting as effective cracks of the same Prenormalisation research in environment assisted cracking. In:
Corrosion Standards, II (Edited by P. McIntyre and D. J. Mills),
depth provided that a short crack model is used.
The Institute of Materials, London, pp. 147–160.
However, further work is required to confirm that 14 K. J. Miller and R. Akid (1996) The application of microstruc-
DKth and a0 remain the same for pits and cracks of tural fracture mechanics to various metal surface states. Proc. R.
equivalent depth. Soc. Lond. 1411–1432.
15 R. Akid (1997) The role of stress-assisted localized corrosion
in the development of short cracks. In: Effects of the Environment
Acknowledgement on the Initiation of Crack Growth (Edited by W. A. Van Der
Sluys, R. S. Piascik and R. Zawierucha), ASTM STP 1298,
This work was carried out as part of the ‘Environmental
American Society of Testing and Materials, pp. 3–17.
Degradation of Materials Programme’, a programme of 16 R. P. Wei (1997) Corrosion fatigue—science and engineering.
underpinning research financed by the United Kingdom Conference on Recent Advances in Corrosion Fatigue, Sheffield, UK,
Department of Trade and Industry. 16–17 April.

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093
 I N F L U E N C E O F P I T T I N G O N FAT I G U E L I F E O F T U R B I N E B L A D E S T E E L 1093

17 K. Toyama and N. Konda (1989) Environmental effects on the 29 Y. Konda and R. Wei (1989) Approach on quantitative evalu-
fatigue crack initiation. In: International Conference on Evaluation ation of corrosion fatigue crack initiation condition. In:
of Materials Performance in Severe Environments, Kobe, Japan, International Conference on Evaluation of Materials Performance in
20–23 November, The Iron and Steel Institute of Japan, Vol. 1, Severe Environments, Kobe, Japan, 20–23 November, The Iron
pp. 105–112. and Steel Institute of Japan, Vol. 1, pp. 135–142.
18 A. I. Lebedeva, V. S. Sokolov, V. F. Rezinskikh and A. F. 30 M. H. Dawling (1977) Crack growth during low cyclic fatigue
Bogachev (1992) The effect of pitting corrosion on the fatigue of smooth axial specimens. In: ASTM STP 637, pp. 97–121.
strength in blade materials. Thermal Engng 39, 69–72. 31 S. Kawai and K. Kasai (1985) Considerations of allowable stress
19 D. C. Buxton, R. A. Cottis, P. A. Scarf and P. McIntyre (1993) of corrosion fatigue (focused on the influence of pitting). Fatigue
Life prediction in corrosion fatigue. In: Corrosion–Deformation Fract. Engng Mater. Struct. 8, 115–127.
Interaction, Fontainebleau, France, 5–7 October, Les Editions 32 S. Ishihara, A. J. McEvily and K. Shiozawa (1995) A fatigue-
de Physique, France, pp. 901–913. crack-growth-based analysis of two-step corrosion fatigue tests.
20 G. Gabetta, L. Torri, C. Rinaldi and F. Bergani (1994) Crack Fatigue Fract. Enging Mater. Strct. 18, 1311–1321.
nucleation and growth from corrosion pits in low pressure rotor 33 S. Suresh and R. O. Ritchie (1984) Propagation of short fatigue
steel. In: Materials for Advanced Power Engineering (Edited by cracks. Int. Metals Rev. 29, 445–476.
D. Coutsouradis, J. H. Davidson, J. Ewald et al.), Liege, Belgum, 34 M. H. El Haddad, T. H. Topper and T. N. Topper (1981)
3–6 October, Kluwer Academic Publishers, 199–207. Fatigue life predictions of smooth and notched specimens based
21 ASTM E466–82 (1994) Standard practice for conducting ampli- on fracture mechanics. ASME J. Engng Mater. Technol. 103,
tude axial fatigue tests of metallic materials. In: 1994 Annual 91–96.
Book of ASTM Standards, Section 3, pp. 465–469. 35 M. H. El Haddad, K. N. Smith and T. H. Topper (1979)
22 R. Qvarfort (1988) New electrochemical cell for pitting cor- Fatigue crack propagation of short cracks. ASME J. Engng
rosion testing. Corrosion Science 28, 135–140. Mater. Technol. 101, 42–46.
23 British Standard (1993) BS 3518, Method of Fatigue Testing. 36 M. H. El Haddad, N. E. Dowling and T. H. Topper (1980)
24 British Standard (1997) BS ISO 4287, Geometric Product J-integral applications for short crack at notches. Int. J. Fracture
Specification (GPS)—Surface Texture—Terms, Definitions and 16, 15–30.
Surface Texture Parameters. 37 S. J. Hudak Jr (1981) Small crack behavior and the prediction
25 R. A. Smith and K. J. Miller (1978) Prediction of fatigue of fatigue life. ASME J. Engng Mater. Technol. 103, 26–35.
regimes in notched components. Int. J. Mech. Sci. 20, 201–206. 38 H. Kitagawa and S. Takahashi (1976) Application of fracture
26 A. D. Cameron and R. A. Smith (1981) Upper and low bounds mechanics to very small surface cracks or the crack in the early
for the lengths of non-propagating cracks. Int. J. Fatigue 3, 9–15. stage. In: Proc. 2nd Int. Conference on Mechanical Behaviour of
27 P. Lukas (1982) Utilization of threshold values for characteriz- Materials, Boston, pp. 627–631.
ation of fatigue notch sensitivity. In: Fatigue Thresholds (Edited 39 H. Dolle and J. B. Cohen (1980) Residual stresses in ground
by J. B. Bäcklund, A. F. Blom and C. J. Bevers), EMAS, Warley, steels. Met. Trans. 11A, 159–164.
UK, pp. 737–758. 40 V. M. Hauk, R. W. M. Oudelhoven and G. J. H. Vaessen (1982)
28 D. L. DuQuesnay, T. H. Topper and M. T. Yu (1986) The The state of residual stress in the near surface region of
effect of notch radius on the fatigue notch factor and the homogeneous and heterogeneous materials after grinding. Met.
propagation of short cracks. In: The Behaviour of Short Fatigue Trans. 13A, 1239–1244.
Cracks (Edited by K. J. Miller and E. R. De Los Rios),
Mechanical Engineering Publications, London, pp. 323–335.

© 1999 Blackwell Science Ltd. Fatigue Fract Engng Mater Struct 22, 1083–1093