Available online at www.sciencedirect.

com
Procedia
Engineering
Procedia
Procedia Engineering
Engineering 00 (2009)
2 (2010) 000–000
441–451
www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Fatigue 2010

Shot-Peening of Steam Turbine Blades: Residual Stresses and their

Modification by Fatigue Cycling
MN Jamesa,b, M Newbya,∗ , DG Hattinghb, A Steuwerb,c
a
Plymouth University, Drake Circus, Plymouth, UK, PL4 8AA
b
Nelson Mandela Metropolitan University, PO Box 77000, Port Elizabeth, South Africa, 6031
c
ESS Scandinavia, Lund University, Box 117, Lund 2210, Sweden

Received 2 March 2010; revised 9 March 2010; accepted 15 March

Abstract

Power generation in thermal stations typically relies on large steam turbines. The corrosion resistant steel blades used in the last
stage of a typical low pressure rotor set are approximately 1m long and experience high centrifugal loading during service. They
operate in a wet steam environment, at approximately 120°C while rotating at 3000 rpm, and failure modes are typically either
stress corrosion cracking or corrosion fatigue. The blades are retained by a fir tree root which is typically shot-peened to
generate compressive residual stresses that resist crack initiation. Finite element (FE) modelling has indicated that, in the
absence of shot-peening, stresses above yield are induced at the fir tree root during operation. To date, no systematic information
has been obtained on the level of residual stresses induced in the fir tree by shot-peening and their relaxation, nor are there any
guidelines as to the magnitude of residual stresses necessary to ensure integrity of the turbine over a life span of at least twenty
years. At least one of these blades has failed in recent years causing a catastrophic failure and severe damage to the turbine-
generator set.
This paper will report results from a comprehensive program of residual stress measurements at the shot-peened fir tree roots of
service blades, and in specimens that simulate the root, using diffraction data from laboratory and synchrotron X-ray radiation
(SXRD). Shot-peening coverage between 75% and 200% was used and stresses were measured up to 5 mm into the
blades/specimens. Measurements were made in the as-peened condition and after applying cyclic stresses representative of
overspeed proof testing and service operation. The results will be used to calibrate FE modelling of residual stresses and as input
into fatigue life prediction.
c 2010 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Keywords: X-ray diffraction; residual stresses; turbine blades; fatigue; shot-peening

1. Introduction

Power generation in thermal stations relies on large steam turbines; a typical low pressure (LP) rotor set is shown
in Figure 1. The last-row blades are approximately 1m long made from 12Cr corrosion resisting steel and experience
high centrifugal loading during service. In service the blades are highly stressed components that operate in a wet

∗
Mark Newby. Tel.: +27-11-629-5192; fax: +27-86-664-8570.
E-mail address: [email protected].

1877-7058 c 2010 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2010.03.048
442 M.N. James et al. / Procedia Engineering 2 (2010) 441–451

2 M Newby et al. / Procedia Engineering 00 (2010) 000–000

environment, with temperatures of approximately 120°C, at 3000 rpm where the failure modes most often
encountered are stress corrosion cracking or corrosion fatigue [1]. They are retained by a fir tree root, see Figure 2,
which is typically shot-peened to resist crack initiation through the existence of compressive residual stresses at the
surface.
Shot-peening is commonly used to improve the fatigue life of metallic components [2]. The process consists of
exposing the surface of the component to a stream of small steel shot by means of compressed air. The kinetic
energy of the shot impacting on the surface of the component causes a thin layer of plastic strain to take place. The
constraint of the structure beneath the plastic layer then forces the surface into a compressive stress state. This
surface effect slows down crack initiation significantly and thus improves fatigue life.
There are some concerns over whether the shot-peening is effective in this application, i.e. around the fir-tree
area [3]. Over-speed tests, up to 3300 rpm, are conducted periodically, normally after maintenance that may affect
the balance of the rotor. This activity can result in localised stresses causing yielding of the material, which may
reduce the level of compressive stress. In addition the varying profiles of the fir tree configuration make the
application of uniform shot-peening difficult which gives rise to some uncertainty about the effectiveness.
It is worth mentioning in this context that the fracture in the fir tree of one of these blades at a South African
power station in January 2003 caused a catastrophic failure and this resulted in a fire, severe damage to the turbine-
generator set and loss of production of a 600MW unit for a period of six month with total damages amounting to
nearly 100 M EURO.
Until recently it has not been possible to measure the residual stresses effectively in a typical blade attachment
profile due to the complex shape and nature of the material. Recent work using high energy, high intensity X-rays
generated by a synchrotron radiation source (SXRD) has shown that residual stress measurements with a high spatial
resolution and reasonable depth are possible in steels[4]. Building on this expertise a systematic investigation was
planned and executed at the European Synchrotron Radiation Facility (ESRF) located in Grenoble, France. The
aims of the present investigation were:
• To quantify optimum shot-peening conditions on flat plate specimens in terms of the residual stress
magnitude and distribution.
• To apply over-speed/fatigue stress cycles to the flat plate samples in-situ on a beamline to simulate the
service fatigue effects.
• To compare the residual stress profiles of the flat samples to those of the fir-tree samples.
These tests will thus allow the optimum shot-peening conditions to be extrapolated to real blades in service. This
work is extremely important for the steam turbine industry and forms part of a research project that is supported by
industry, allowing its findings to have a real impact on industrial practice.

Fracture
plane

Figure 1: Typical LP Turbine Rotor Figure 2: Blade, Fir tree root and Fractured root section
 M.N. James et al. / Procedia Engineering 2 (2010) 441–451 443

M Newby et al. / Procedia Engineering 00 (2010) 000–000 3

2. Experimental Procedure

2.1. Sample Preparation

A number of samples were prepared; all taken from a single used turbine blade to ensure minimal variation in
material properties. Twelve flat hour glass samples were machined as shown in Figure 4, the reduced cross section
was 5×5mm. Three fir tree samples were machined as shown in Figure 3. Three round tensile test samples, 5 mm
diameter, were also machined. Mechanical tests were conducted on the round samples to determine the Elastic
Modulus, Yield and Ultimate Stresses. The material that was used to manufacture the samples was a 12CrNiMo
martensitic steel with a Din No 1.4939, and a trade name of Jethete. This material is a creep and corrosion resistant,
hardened and tempered steel with a usual upper temperature limit of 560 °C. It has exceptional toughness and creep
rupture strength. The average chemical composition from the data sheet is shown in Table 1, and the average of the
three mechanical tests is shown in Table 2.

Table 1: Chemical composition [%]

C Si Mn Cr Mo Ni V N
0.12 0.20 0.80 11.70 1.70 2.70 0.30 0.04

Table 2: Mechanical properties (at room temperature)

0,2% Proof Stress UTS Elastic Modulus Poisson’s ratio

[MPa] [MPa] [GPa]
868.1 1047.7 204.2 0.3

The fir tree and flat samples were all machined using wire cutting, and then ground, to minimise machining
stresses. The samples were shot-peened at the South African Airways technical services division located OR Tambo
airport in Johannesburg. Most of the turbine blades used in South African power stations have been treated at this
facility. A guideline [5] developed by Eskom (the national power utility in South Africa) and one of the original
equipment manufacturers was used in determining the parameters for the shot-peening of the flat samples. The
samples were subjected to four different coverage conditions (75%, 100%, 150%, 200%). This resulted in three
samples for each condition. The samples were shot-peened from both sides simultaneously to prevent any bending
of the sample and to allow for symmetrical analysis during the diffraction measurements.
Three fir tree samples were made and mounted in an old blade as shown in Figure 3, in order to get conditions
that were exactly the same as those experienced by a normal
blade root. As per the procedure the coverage in this case was Table 3: Shot-Peening Parameters [5]
200%.
PARAMETER COMMENT
The Eskom guideline calls for six nozzles to be used during
the shot-peening process. The nozzles were mounted at Intensity: 8A-12A
specific angles and distances from the blade surface. The Air Pressure: Set to achieve Almen curve
blade was mounted on a turntable and as this rotated, the Turn Table Speed: 50v (10 rpm)
blade root area travelled through the nozzle matrix. The Exposure Time: Set to achieve Almen curve
nozzles moved up and down vertically during this process.
Nozzle Size: 5/16”
When the system was being set up, a test blade with place for
No. of Nozzles: 6
seven Almen strips was mounted on the turntable and the time
required to achieve 100% coverage was measured. This time Lance travel: 145mm (6.150”)
was then doubled to obtain 200% coverage. The parameters Lance cycles: 1.5 cycles for set distance
are shown in Table 3. Shot- Size: CW28 = S-230 (or smaller)
Coverage: 200%
444 M.N. James et al. / Procedia Engineering 2 (2010) 441–451

4 M Newby et al. / Procedia Engineering 00 (2010) 000–000

2.2. Measurement of Residual Stress

Residual stresses can be measured in a variety of ways, some of

which are at least semi-destructive. Measuring the effects of shot-
peening requires a method that is non destructive and that is able to
establish the profile with depth and high spatial resolution over the first
0.5mm. SXRD provides this capability, however access to this type of
equipment can be difficult to obtain. Laboratory X-ray techniques can
be used to complement the synchrotron experiments but measurement
with depth requires layer removal.

The basics of diffraction stress measurements are covered

extensively by Withers and Bhadeshia [6] but are repeated here for the
sake of completeness. The basic concept relies on using the crystalline
lattice as an atomic strain gauge. Changes in the lattice spacing, due to
the local stress, either extensions or reductions can be measured by
changes in the diffraction peak positions. Diffraction techniques can
thus measure the strain tensor in the material and the stress may be Figure 3: Three fir tree samples after shot-peening
inferred using the material constants. The changes in the diffracted and showing the sectioned test pieces
angle are defined by Bragg’s equation;

λ = 2d hkl sin θ (1)

where; Ȝ = incident wave length, dhkl = lattice plane spacing of a particular hkl reflection and ș = the diffraction
angle.

The strain, in its generic form for all diffraction planes, may be calculated from the difference between the
measured spacing d and the strain free lattice spacing d0, divided by d0. As the material was assumed to be
elastically isotropic, hence for the sake of brevity we write dhkl = d.

d − d0
ε= (2)
d0

This assumes that the d0 value is known, either locally e.g. across a weld or globally, which is often not the case.
In diffraction measurements the d0 value can be determined or estimated in a number of ways; using powder
samples, stress and moment balancing, annealed reference samples or choosing an area that is as stress free as
possible [7].
In laboratory XRD measurements (and likewise in SXRD) the problem of unknown reference d0 can be
overcome if the material properties are known so that the basic X-ray elastic constants can be calculated, and the
sin²ȥ technique is used to analyse the data[7,8]. In the absence of texture or strong stress gradients, the sin²ȥ
technique relies on a linear relationship between the lattice spacing and the sin² value of the angle of inclination (ȥ)
of the sample. Approximately ten data points should be recorded over a change in ȥ of about 70°. The slope of the
graph d vs sin²ȥ , combined with the X-ray elastic constants will enable the stress to be calculated independent of
the unstrained lattice parameter.

For an elastically isotropic material the X-ray elastic constants S1 and S2 may be calculated from the bulk elastic
constants as shown in equations 3 and 4 [9];

−ν 1 1 +ν
S1 = S2 = (3,4)
E 2 E
 M.N. James et al. / Procedia Engineering 2 (2010) 441–451 445

M Newby et al. / Procedia Engineering 00 (2010) 000–000 5

For an equi-biaxial stress state the stress may be calculated from either the transverse or longitudinal strain
component as shown in equations 5 and 6 [10];

E( 1 + ν ) − Eε trans
σ long = ε long or σ long = (5,6)
( 1 −ν 2 ) ν

where; E = Elastic modulus , Ȟ = Poisson’s ratio , ılong = longitudinal stress , İlong = longitudinal strain and
İtrans = transverse strain. This set of equations only applied to the samples prior to fatiguing, which renders the
residual stress system biaxial, and the conventional expressions for biaxial stress systems apply, and have been
employed, for the samples in that case.

2.3. Laboratory XRD measurements

In order to check the consistency of the shot-peening process a piece was cut off the end of each flat sample and
analysed using laboratory XRD analysis. The samples had to be polished mechanically so that the surface was
smooth enough for a reliable measurement. This had to be done without introducing additional stress. The aim was
to polish down to a 3μm finish, the penetration of the beam was estimated to be 18μm. An initial layer was
removed with 1200 grit water paper and then 3 μm diamond paste was used.
The samples were analysed using a PANalytical X’pert Pro X-ray diffraction machine at Pretoria University.
The beam was focused through slits 1mm wide and 0.25mm high. This resulted in an irradiated area of 1.7mm² at ȥ
=0° and 4.7mm² at ȥ=71°.
The samples were mounted in a holder and scans
were recorded on two orthogonal directions. A Table 4: XRD results after shot-peening
cobalt source was selected with a characteristic
wavelength of 1.7890111Å. The 211 plane was SAMPLE SHOT-PEENING STRESS STD DEV
used to give as high a 2ș value as possible resulting NUMBER CONDITION [MPa] [MPa]
in a peak around 99.2°. Peaks measured at higher 1 75% -568.5 19.9
2ș values will exhibit a larger shift due to residual
2 75% -549.6 18.8
stress and thus provide more accurate result [8].
A W-powder sample was used to calibrate the 3 75% -576.5 18.4
machine geometry. The vertical height of the 4 100% -588.9 13.5
sample was adjusted until the reference stress was 5 100% -572.8 12.8
as close as possible to zero. This datum was 6 100% -596.0 9.3
recorded on the dial gauge and used for the shot- 7 150% -552.6 11.7
peened samples. The stress state in the specimens
8 150% -555.9 14.4
was assumed to be equi-biaxial due to the nature of
the shot-peening application. The data was 9 150% -550.1 11.1
processed using the PANalytical proprietary 10 200% -544.0 13.7
software X’Pert Stress. Material constants for the 11 200% -576.0 14.1
samples as determined by the mechanical tests were 12 200% -564.3 14.9
entered into the database and the X-ray elastic
constants calculated. The regression fit was generally very good on the sin²ȥ plots. The results for the twelve
samples are summarised in Table 4.

2.4. Synchrotron XRD Measurements

The synchrotron measurements were conducted at the ESRF on beam line ID31, experiment MA326, in
monochromatic mode at 60 keV, and later on ID15A, experiment ME1165, in energy dispersive mode with energies
up to several hundred keV. In both cases the high energy X-ray provided enough penetration capabilities to work in
transmission geometry for the longitudinal and transverse direction in the 5mm centre of the specimen. The beam
dimensions on ID31 were set to 50 and 300 microns. The diffraction angle was around 10 deg, resulting in a
446 M.N. James et al. / Procedia Engineering 2 (2010) 441–451

6 M Newby et al. / Procedia Engineering 00 (2010) 000–000

diamond-shaped, elongated gauge volume. The samples for ID31 were orientated so that the strain could be
measured first in the longitudinal direction and then in the transverse direction (with a corresponding change in the
beam dimensions). The samples are shown in Figures 4 and 5. It was not possible to get a third (short transverse)
direction due to the length of the samples, and the corresponding change in gauge volume configuration.. The
stresses were assumed to be equi-biaxial, as was the case for the laboratory XRD measurements. This meant that
the longitudinal stress could be calculated from either the longitudinal or the transverse strain [10] individually,
providing a means of cross-checking the assumption as well as the results.
The specimens investigated consisted of;
• Two specimens from each condition were analysed in the initial shot-peened condition.
• One set was loaded to 868 MPa (0.2% proof stress), and the other to 600 MPa and then re-tested.
• One reduced set was loaded to 910 MPa (0.5%), the other to 100000 cycles at 868 ±10 MPa then re-tested.
• The fir tree samples were only measured in the transverse direction.
In each case the machine was programmed to take scans from the surface at 20μm for ten steps, then 50μm for
six steps and then 1mm steps across the specimen. The data from ID31 was reduced to a lattice spacing value d0 for
each scan by using the LAMP software supplied by the ILL and adapted to ID31 by the FaME38 facility.

Strain measured

Diffracted Incident beam

beam

Figure 4: Transverse scan set-up Figure 5: Fir-tree scan set-up illustrating diffraction geometry
for normal strain measurement

The data shown in Figures 7 and 8 has been taken from the experiment on ID31, and represents the condition
after shot-peening and before any loading was applied, for all specimens. An average d0 value was calculated by
using a stress balancing calculation in the longitudinal direction. This was possible because the synchrotron beam
could penetrate right through the 5mm sample and data points could be gathered with sufficient resolution to
characterise the stress profile.

The fatigue loading was re-done on beam line ID15A at a later date using Sample 6, which had been subjected to
a 100% peening coverage, and had not been
used during the ID31 tests. The maximum Table 5: Load matrix for Sample 6
load was limited to the 0.2% proof stress
value of 868 MPa. The dynamic stress LOAD NUMBER OF CYCLES
range was based on a conservative estimate 620 MPa 1 cycle
of the forces experienced by a turbine blade 600 ± 20 MPa 10, 100, 1000, 10000 and 100000 cycles R=0.935, f=5Hz
during normal operation. The following
868 MPa 1 cycle
(see Table 5) load matrix was applied;
848 ± 20 MPa 100 and 10000 cycles R=0.953, f=5Hz
 M.N. James et al. / Procedia Engineering 2 (2010) 441–451 447

M Newby et al. / Procedia Engineering 00 (2010) 000–000 7

Combined Values - After Shot Peening (Zoom)

200

-200
Stress [MPa]

-400

-600

-800

-1000
0 1 2 3 4 5

Displacement [mm]

S1-75% S2-75% S4-100% S5-100% S7-150% S8-150% S10-200% S11-200%

Figure 6: Longitudinal stress for all samples after shot-peening

Figure 6 shows data from eight samples, two from each coverage condition. Figure 7 shows the data over the
first 0.5 mm depth into the sample. The stress profile is consistent with published data showing a compressive stress
reaching a maximum just below the surface, with the trend going tensile approximately 0.25 mm into the depth of
the samples. There was no significant difference between the four different shot-peening conditions. Figure 7
illustrates that the highest compressive stress occurred on Sample 2 with 75% coverage. There were some
differences in the depth of penetration, as expected the greatest depth was on a sample with 200% coverage.
The measurements after the 0.2% proof load of 868 MPa showed that the residual stress had reduced
significantly, which was to be expected as some yielding would occur. The fatigue loading at 910 MPa resulted in
the stress being relaxed almost immediately and thus analysis of additional cycles was unnecessary.
The system errors in peak detection were very small compared to the material point to point variation. This error
is estimated to be approximately 15 MPa.

Combined Values - After Shot Peening (Zoom)

200

-200
Stress [MPa]

-400

-600

-800

-1000
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

Displacement [mm]

S1-75% S2-75% S4-100% S5-100% S7-150% S8-150% S10-200% S11-200%

Figure 7: Residual stress profile over first 0.5 mm for all samples
448 M.N. James et al. / Procedia Engineering 2 (2010) 441–451

8 M Newby et al. / Procedia Engineering 00 (2010) 000–000

Fir Tree Profile

200

100

-100
Stress [MPa]

-200

-300

-400

-500

-600

-700

-800
0 0.5 1 1.5 2 2.5 3 3.5 4

Displacement [mm]

Figure 8: Residual stress in the top serration of the fir-tree (determined from transverse strain sensor)

The fir tree sample configuration did not allow for the measurement of the longitudinal strain tensor, so only a
transverse measurement was done. The stress profile for the top serration of the fir tree is shown in Figure 8. The
depth penetration on the fir tree samples was significantly more than on the flat samples, the zero crossing point
moved from 0.22mm to 0.45mm, probably due to the concentration of shot-peening impacted from the multiple
nozzle configuration.

Sample 6 - Fatigue loading at 600 ±20 MPa plus 868MPa cycle

1000

500 As peened

1 cycle
0

10 cycle
-500
Strain [µm/m]

100 cycles
-1000
1000 cycles
-1500
10000 cycles

-2000
100000
cycles
-2500 868 Mpa

-3000

-3500
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Distance [mm]

Figure 9: Residual stress profiles for different fatigue cycles on Sample 6

A summary of the fatigue data for the tests with a mean load of 600 MPa is shown in Figure 9 and 10. The
results in Figure 9, which has measurement data for the whole cross-section, show that the fatigue loading up to
100000 cycles had no discernable effect on the residual stress profiles. Figure 10 shows the results in detail for
sample 6. This is further illustrated in Figure 11 which shows the data over the last 0.5 mm of the sample width. A
third order polynomial fit has been applied to the data, and it is clear that the stress variation is very small.
 M.N. James et al. / Procedia Engineering 2 (2010) 441–451 449

M Newby et al. / Procedia Engineering 00 (2010) 000–000 9

Sample 6 - Fatigue loading at 600 ±20 MPa - Zoom

200
As Peened
100
1 Cycle

0 10 Cycles

-100 100 Cycles

1000 Cycles
-200
Stress [MPa]

10000 Cycles
-300
100000 Cycles
-400
Poly. (As Peened)
-500
Poly. (1 Cycle)
-600 Poly. (10 Cycles)

-700 Poly. (100 Cycles)

-800 Poly. (1000 Cycles)

Poly. (10000 Cycles)

-900
4.7 4.75 4.8 4.85 4.9 4.95 5 5.05 Poly. (100000 Cycles)
Distance [mm]

Figure 10: Detail of the last 0.25mm of the residual stress profiles

Sample 6 - Fatigue loading at 868 ±20 MPa - Zoom

0

-50
1 CYCLE

-100
100 CYCLE
Stress [MPa]

10000 CYCLE
-150

Poly. (1 CYCLE)
-200
Poly. (100 CYCLE)

-250 Poly. (10000 CYCLE)

-300

-350
4.75 4.8 4.85 4.9 4.95 5

Displacement [mm]

Figure 11: Residual stress profiles after fatigue loading at 0.2% proof stress

The next stage in the fatigue testing was the application of loading up to 868 MPa. The first cycle is again shown
in Figure 11, together with the data for 100 and 10000 cycles. Experimental time limited the data collection of
further load cycles. The results shown in Figure 11 have been calculated from the longitudinal strain tensor, but the
assumption of an equi-biaxial stress state no longer applies due to the yielding that has taken place. A comparative
stress calculation has been done to trend the reduction in residual stress. Data for the strain tensor at 45° to the
longitudinal direction was collected for the 10000 cycle test, but is not shown in this paper.
Figure 11 shows the measurements over the last 0.25 mm of the sample and again a third order polynomial fit has
been applied to the data. The reduction in residual stress is clearly evident as an effect of the fatigue loading. The
polynomial curves were used to calculate the stresses at a depth of 50 microns and these values have been plotted in
Figure 12. The data shows a log-linear trend up to 10000 cycles and an extrapolation has been done to 100000
cycles as an illustration of the reduction in residual stress if this trend continued.
450 M.N. James et al. / Procedia Engineering 2 (2010) 441–451

10 M Newby et al. / Procedia Engineering 00 (2010) 000–000

Reduction in Residual Stress with Fatigue

-50
Extrapolation of data to 10E7
cycles assuming log-linear
behaviour
-100
Logarithmic regression fit to
fatigue data. R² = 0.9833
Stress [MPa]

-150

-200

-250

-300

-350
1 10 100 1000 10000 100000 1000000 10000000

Number of Cycles

Stress 50 microns below surface) Log. (Stress 50 microns below surface))

Figure 12: Illustration of the reduction in residual stress after fatigue loading at 0.2% proof stress

3. Conclusions

The Synchrotron XRD results provided clear profiles of the residual stresses through the 5mm cross section of
the samples at high spatial resolution with the capabilities of undertaking in-situ fatiguing. The results showed the
typical profiles expected from shot-peening with the maximum compressive stress just below the surface, gradually
changing towards tension at a few hundred microns. There was no discernable effect from the fatigue loading when
the mean stress was set at 600 MPa. When the mean stress was set to 20 MPa below the 0.2% proof stress of 868
MPa there was a definite effect on the residual stress with the compressive peak value having a log-linear trend up to
10000 cycles. The comparison to FE modeling and full dissemination of the results is beyond the scope of this paper
and will be presented elsewhere.

Acknowledgements

The ESRF for beam time allocation, and assistance from Drs Thomas Buslaps and Alex Evans during
experiments ME326 and ME1165. The Eskom research program for funding allocations. Dr Sabine Verryn from
the University of Pretoria, as well as the South African Airways technical services.

References

[1] Pillay ., Stress Corrosion Cracking. In Eskom Power Plants - Scan Report, TRR/S96/219, 1996
[2] Marsh KJ, editor. Shot peening: techiques and applications. London, UK: EMAS; 1993.
[3] Nel W. Duvha Power Station – blade analysis. Eskom – Technology Services International Reference : MT320/2005, 2005
[4] Steuwer A, Santisteban JR, Turski M, Withers PJ, Buslaps T. High-resolution strain mapping in bulk samples using full-profile analysis of
energy-dispersive synchrotron X-ray diffraction data. Journal of Applied Crystallography 2004;37:883.
[5] Botha JP, SAA Technical procedure, Shot peening process for stage 5 turbine blades, Rotek and Eskom
[6] Residual Stresses: Measurement by Diffraction, Encyclopaedia of Materials: Science and Technology, pp 8158-8170 ISBN: 0-08-
0431526, 2001
[7] Withers PJ, Preuss M, Steuwer A, Pang, JWL, Methods for obtaining the strain-free lattice parameter when using diffraction to determine
residual stress. J. Appl. Cryst. 40, 891-904, 2007
[8] Hauk V and Krüger B. Invariant intersections in diffraction stress analysis — a check of experimental results, Materials Science and
Engineering A 284 pp 261–267, 2000
 M.N. James et al. / Procedia Engineering 2 (2010) 441–451 451

M Newby et al. / Procedia Engineering 00 (2010) 000–000 11

[9] Vermeulen AC, An Elastic constants database and xec calculator for use in xrd residual stress analysis, JCPDS-International Centre for
Diffraction Data 2001,Advances in X-ray Analysis, 44 pp 128-133, 2001.
[10] Hearn EJ, Mechanics of Materials Volume 1 and 2, ISBN 0080311318 V.1 and V.2, 1985.
[11]PANalytical, XRD Course Residual Stress, PANalytical B.V., The Netherlands, 2006