You are on page 1of 10

Proceedings of the 2012 9th International Pipeline Conference IPC2012 September 24-28, 2012, Calgary, Alberta, Canada

IPC2012-90701
X-RAY DIFFRACTION STUDY OF MICROSTRUCTURAL CHANGES DURING FATIGUE DAMAGE INITIATION IN STEEL PIPES
Bianca Pinheiro Laboratory of Mechanics of Lille University Lille 1 Lille, FRANCE Jacky Lesage Laboratory of Mechanics of Lille University Lille 1 Lille, FRANCE

Ilson Pasqualino Subsea Technology Laboratory COPPE - Federal University of Rio de Janeiro Rio de Janeiro, BRAZIL

Noureddine Benseddiq Laboratory of Mechanics of Lille University Lille 1 Lille, FRANCE

Edoardo Bemporad University of Rome "ROMA TRE" Mechanical and Industrial Eng. Dept., Via Vasca Navale 79, 00146 Rome, ITALY

ABSTRACT The present work aims to evaluate the microstructural mechanisms associated with the initiation of fatigue damage in steels used in the petroleum industry. Microdeformations and residual stresses (macrostresses) are evaluated by X-ray diffraction in real time during alternating bending fatigue tests performed on samples taken from an API 5L X60 grade steel pipe. Microdeformations are evaluated from measurements of the full width at half maximum (FWHM) of the diffraction peak and residual stresses are estimated from the peak displacement. The evolution of microdeformations shows three regular successive stages of changes. The amplitude of variation of each stage is intensified with increasing stress amplitude, while the duration is reduced. A similar evolution is found for residual stresses, whose stages of changes have nearly the same durations as those of microdeformations. Changes in density and distribution of dislocations are observed by transmission electron microscopy combined with the technique of focused ion beam. To understand the role of the initial structure, fatigue tests on annealed samples are performed under the same test conditions. Again, three stages of changes are observed, but with an increase in microdeformations instead of a decrease during the first stage due to the initial state of the dislocation network. The results are very encouraging for the consideration of microstructural changes measured by X-ray diffraction in the development of a future indicator of fatigue damage initiation in API 5L X60 grade steel pipes.

INTRODUCTION Cyclic loads undergone by steel pipes used for oil and gas exploitation can lead to their fatigue failure. As a matter of fact, fatigue is one of the major causes of failure of these structures [14], which can be followed by catastrophic environmental damage and also significant financial loss. To predict a fatigue failure and assure the structural integrity of oil and gas pipes, usually made of high strength steels, such as API 5L X60, X65, X70 and X80 grade steels [5], it is important to be able to detect and follow fatigue damage during operation. The fatigue damage process may be split in two phases: an incubation phase, during which only microstructural changes, microcracking and microcrack nucleation can be observed, and a propagation phase, characterized by macroscopic cracking and macrocrack propagation that lead to fatigue failure. In the second phase, fatigue damage (macrocrack propagation) is easier to be detected and followed. Moreover, it can be modeled by the classic Paris law [6] and modified versions within the linear elastic fracture mechanics (LEFM) approach. Besides that, risk-based inspections are usually based on the probability of macrocrack detection, as in the case of steel pipelines [7]. Regarding the first phase, even if the mechanisms developed in early stages of fatigue are well known multiplication of dislocations, formation of slip bands, extrusions, intrusions, and microcracks [810] it is not always clear which microstructural changes, and to what extent, can be associated with fatigue damage. Another difficulty concerning this phase is related to the observation techniques employed. Since damage evolution, at least at the onset of fatigue, is concerned

Copyright 2012 by ASME

by the movement, density and arrangement of dislocations, changes have to be observed at the nanoscale. However, the destructive character of the observation techniques available for this purpose renders them inconvenient, since in this case it is not possible to perform a series of observations on a same sample during fatigue test. In addition, they require the extraction of very small samples or even thin lamellae from the test piece, which involves complex and very time-consuming technical procedures. Therefore, the use of nondestructive evaluation (NDE) techniques to investigate physical and microstructural changes related to fatigue has greatly increased in the recent past. These techniques include thermography [11], ultrasonic testing [12,13], magnetic inspection [14,15], or X-ray diffraction [16 19]. Although most of the currently used NDE techniques may help to detect microstructural changes during the damage process, it is still not possible to correlate fatigue damage to these changes for damage quantification and prediction of the residual life of components submitted to cyclic loads. Among available NDE techniques, though, X-ray diffraction (XRD) offers the more interesting perspectives, since it can deliver relevant information about the dislocation network state and microstructural changes. The aim of this work is to evaluate and quantify microstructural changes in real time during fatigue initiation and provide ground work for the development of an indicator of fatigue damage initiation that could allow the residual life prediction of steel pipes submitted to cyclic loads, before macroscopic cracking. The X-ray diffraction technique is used to evaluate microdeformations, characterized by the full width at half maximum (FWHM) of the XRD peak, and macro residual stresses in real time during high cycle fatigue (HCF) tests on API 5L X60 grade steel samples under alternating bending loads. Fatigue tests are conducted at five different stress amplitudes, with at least three tests for each load level. Microstructural changes in terms of variations in FWHM and residual stresses are evaluated during fatigue cycling as a function of the alternating stress level applied. Samples fatigued to specific numbers of cycles are further examined by transmission electron microscopy (TEM) to investigate the correlation between FWHM changes and dislocation density. In addition, fatigue tests on annealed samples are also carried out to study the effects of restoration of the nonstressed dislocation network. The experimental results obtained are analyzed in view of the development of an indicator of fatigue damage initiation for the X60 grade steel that could allow the evaluation of the residual life before macroscopic cracking and help to increase the reliability of pipelines subjected to cyclic loads. X-RAY DIFFRACTION TECHNIQUE FOR FATIGUE CHARACTERIZATION The X-ray diffraction (XRD) technique presents two main interests for the study of microstructural changes during fatigue. Considering that X-ray penetration depth reaches about 510 m, measurements are restricted to the near-surface zone of the material and the technique is particularly suitable, since during the fatigue process major microstructural evolutions take place in this zone. Additionally, this technique is nondestructive and can be employed several times during a fatigue test. The

use of modern portable systems extends the feasibility of the technique, allowing direct evaluation of the surface of samples in real time during testing. The shape and position of an XRD peak depend on microstructural parameters such as crystallite size, residual stresses, microdeformations, stacking faults, etc. [20]. It is well known that the analysis of broadened XRD peaks can be used to study microstructural changes of plastically deformed crystalline materials [21]. Schematically, a material may be considered as a multitude of regular piling of atomic planes bounded by a dislocation network, called coherently diffracting domain (CDD) and characterized by its lattice spacing d0 (nonstressed material). Changes in lattice spacing arise from deformation of the crystal lattice space, which can be detected by XRD, as illustrated in Fig. 1. The deformation of the crystal lattice space produces different effects on the XRD peak whether it is classified as uniform or nonuniform. In the case of a uniform deformation = d/d0 ( < 0.2%), first order stresses (macrostresses) are concerned, being homogeneous over many grains, i.e., comprising few hundreds of microns. In this case, the XRD peak corresponding to a family of crystalline planes is shifted, following the Braggs law, and macrostresses can be estimated by the sin method [22]. This is schematically shown in Fig. 1(a). A nonuniform deformation is related to second and third order stresses (microstresses), which are homogeneous over small domains on the order of tens of microns, such as a part of a grain. When microstresses (microdeformations) are concerned, broadening of the XRD peak is observed (Fig. 1(b)), related to changes in the dislocation density [8]. XRD peak broadening can be studied by peak profile analysis, based on mathematical formulae, which can estimate microdeformations and CDD sizes [21,2325]. A very common parameter used to characterize peak broadening is the full width at half maximum (FWHM) [1721,2628], which can give at least qualitative information about the dislocation network state.

Figure 1. Influence of (a) macrostresses (uniform deformation) and (b) microdeformations (nonuniform deformation) on X-ray diffraction peak.

MATERIAL PROPERTIES CHARACTERIZATION The evolution of fatigue damage initiation was investigated on API 5L X60 grade steel samples under two different conditions: as-machined (hardened) and annealed. The

Copyright 2012 by ASME

annealing treatment was employed to study the effects of restoration of the dislocation network state affected by the pipe manufacturing process. Annealed samples were heated at 850C for 1h under protective atmosphere and then cooled in the furnace. Material properties of X60 grade steel were characterized by means of chemical composition analyses, metallographic analyses, uniaxial tension tests, and Vickers microhardness tests [29,30]. A pearlitic-ferritic microstructure was observed through metallographic analyses [29,30]. Uniaxial tension tests were conducted in a conventional servo-hydraulic machine (Instron 8802) with coupons machined in the longitudinal direction of an X60 grade steel pipe sample and deformed at a strain rate around 1.710-4 m/ms-1 [29,30]. Table 1 presents the average mechanical properties obtained, which are in agreement with the standard API SPEC 5L Specification for Line Pipe [5]. Vickers microhardness tests were performed in three different directions, perpendicular to longitudinal, circumferential and radial directions of the pipe sample [29,30]. Table 2 shows average results of Vickers microhardness obtained (longitudinal direction) for X60 grade steel in as-machined (hardened) and annealed conditions. Values of ultimate tensile strength estimated from Vickers microhardness (HV) as Su = 3HV [31 33], shown in Table 2, are very close to the results obtained from uniaxial tension tests (Table 1). Annealing reduced the strength and hardness of the material with respect to the asmachined condition.
Table 1. Average mechanical properties for API 5L X60 grade steel (hardened).

sample failure. XRD measurements were performed after unloading of the sample, and in real time during fatigue tests, i.e., without the need of removing the sample from the testing machine. XRD measurements were taken at the center of the sample gage length, as indicated in Fig. 3, using a computer controlled -diffractometer [22] and employing Cr-K radiation (wavelength of 2.291 ). The incident beam was collimated by a circular aperture with 2 mm diameter, giving an irradiated area of 3.14 mm2. XRD measurements were taken at (211) planes of the ferrite phase (-Fe), under seven tilt angles (0, 3.67, 7.33 and 11), with 3 oscillations. The penetration depth for chromium radiation in ferritic steels is 5.8 m at = 0 [35].

Figure 2. Alternating bending fatigue testing machine [34].

Young modulus (GPa) 183 9

0.2% offset; 50 mm gage length

Yield strength (MPa) 1 520 6

Ultimate tensile strength (MPa) 602 6

Elongation (%) 19 1
2

Table 2. Average results of Vickers microhardness for hardened and annealed API 5L X60 grade steel.

Hardened HV 195 12

Su (MPa) 585 36

Annealed HV 154 7

Su (MPa) 462 21

FATIGUE TESTS API 5L X60 grade steel samples were submitted to high cycle fatigue (HCF) tests under alternating bending loads. Microdeformations as well as macro residual stresses were evaluated during fatigue tests by X-ray diffraction technique. Experimental setup The experimental setup for fatigue tests consisted of a fatigue testing machine Schenck model PWON [34], schematically illustrated in Fig. 2, and a diffractometer Proto iXRD (portable mode) [29,30]. Samples were submitted to room temperature HCF tests under strain-controlled alternating bending loads (sinusoidal waveform) at a frequency of about 25 Hz. The dual eccentric of the fatigue testing machine was used to set up the load levels applied (Fig. 2). Fatigue tests were regularly interrupted for XRD measurements at periods of 10,000 load cycles maximum up to

Figure 3. Experimental setup for X-ray diffraction measurements (focal distance F = 40 mm).

Figure 4. Geometry and dimensions (in mm) of fatigue test samples (Thickness: 2.90 mm).

Copyright 2012 by ASME

Table 3. Fatigue test results of as-machined samples.

Sample T1EF03 T1EF09 T1EF10 T1EF04 T1EF05 T1EF08 T1EF15 T1EF11 T1EF12 T1EF13 T1EF22 T1EF23 T1EF24 T1EF06 T1EF07 T1EF14
1

Strain Stress amplitude amplitude (MPa) 0.15% 277 / 277


1

0.53 o -0.8

0.17%

319 / 319

0.61 o -1.1

0.18%

340 / 348

0.65 o -1.7

0.19%

361 / 361

0.69 o -1

0.19%

367 / 401

0.71 o -2.8

N (cycles) 296,200 306,600 288,300 147,300 106,600 143,200 191,100 56,300 59,400 62,700 63,600 190,400 48,100 60,200 59,300 41,700

Fatigue test samples Fatigue test samples were machined from the same pipe sample used to produce test coupons for material properties characterization [29,30]. The geometry and dimensions of fatigue test samples are presented in Fig. 4. Fatigue test samples were prepared under two different conditions: asmachined and annealed. In both conditions, samples were electrolytically polished in order to reduce surface roughness. Surface roughness can affect XRD measurements since it reduces the density of diffracting crystallographic planes, leading to lower precision of diffraction peaks. Through removal of a thin surface layer of about 1 m depth, electrolytic polishing can help to reduce surface roughness and, consequently, to increase the accuracy of XRD measurements. In addition, electrolytic polishing can attenuate machininginduced surface residual stresses. Electrolytic polishing was done within the gage length region of the sample surface with the aid of a perchloric-based acid solution (standard electrolyte Struers A2) and an adapted mask. The average roughness Ra was reduced from 0.405 to 0.206 m after electrolytic polishing in as-machined samples, which was followed by a reduction in average residual stresses from 6211 to 456 MPa [29,30]. RESULTS AND DISCUSSION

Corrected by Gerber mean stress criterion


900 800 700

Exp. results (as-machined samples) Basquin equation (exp. results) Sn = 1955 N


-0.1520

600 500 400 300

Exp. results [C.S. Vianna, 2005]

200

100 4 10

10

10

As-machined samples As-machined samples were submitted to fatigue tests at five different alternating bending loads, with four loads applied under stress ratios R around -1 (fully reversed stresses) and one under R = -2.8, at the highest strain amplitude. At least three fatigue tests were carried out for each alternating bending load. The obtained fatigue test results are presented in Table 3. Fatigue lives were determined at the rupture of samples (after complete sample separation). Fatigue test results were corrected according to the Gerber mean stress criterion [36] to bring every result to the condition of fully reversed stresses (R = -1). Fig. 5 shows average fatigue test results obtained for each load level applied. A non-statistically evaluated Basquin equation was obtained from fatigue test results (Fig. 5), given as
S n = 1955 N 0.1520

Alternating stress amplitude (MPa)

N (cycles)

Figure 5. S-N diagram of fatigue test results of as-machined samples.

(1)

Signals from two X-ray detectors were transferred to a data acquisition system controlled by computational software. A Gaussian distribution function was used to fit XRD peaks, and then microdeformations, characterized by FWHM, and macro residual stresses were estimated. Macro residual stresses were calculated according to the sin method [22], considering the lattice spacing in the nonstressed material d0 = 1.1702 , diffraction angle 2 of 156.4, and elastic constants S1(211) = v/E = 1.28106 MPa1 and S2(211) = (1+)/E = 5.92106 MPa1 for -Fe phase (211) planes. FWHM values and macro residual stresses were considered in sample longitudinal direction. Uncertainties were lower than 7% in measurements of FWHM values and on the order of 10 MPa in residual stress measurements.

with Sn in MPa. The fatigue test results are in good agreement with test results previously obtained for the same material under identical test conditions [37] (Fig. 5). In Figs. 6 to 10 variations in FWHM of the XRD peak measured during fatigue tests at different alternating stress amplitudes are presented. The FWHM change measured at a given number of cycles is defined as the difference between FWHM values at this point and at N = 0, i.e., FWHM FWHM0. For all stress amplitudes, three stages can be identified in FWHM changes during fatigue cycling. The first stage (Stage 1) takes place in the early cycles and is characterized by a fast decrease in FWHM. In the second stage (Stage 2), the rate of decrease in FWHM is considerably reduced. This stage comprises the major fraction of fatigue life (about 50%). Finally, the third stage (Stage 3) occurs in the last cycles with a rapid decrease in FWHM up to the end of fatigue life.

Copyright 2012 by ASME

0.00 -0.05 -0.10

600 500

0.00
a = 361 MPa

600
As-machined samples T1EF22 T1EF24 Average curve

-0.05 -0.10

500 400 300

FWHM - FWHM0
a = 277 MPa

400
As-machined samples T1EF03 T1EF09 T1EF10 Average curve

FWHM - FWHM0

FWHM - FWHM0 ()

FWHM - FWHM0 ()

-0.15 -0.20 -0.25 -0.30 -0.35 -0.40 -0.45 0 40000

300

-0.15 -0.20 -0.25 -0.30 -0.35 -0.40 -0.45 0 10000 20000 30000 40000 50000 60000
R - R0

R - R0 (MPa)

R - R0 (MPa)

200 100 0 -100 -200

200 100 0 -100 -200 -300 70000

R - R0

-300 80000 120000 160000 200000 240000 280000 320000

N (cycles)

Figure 6. Changes in FWHM and R with cycling of asmachined samples at a = 277 MPa (R = -1).
0.00 -0.05 -0.10 600 500

Figure 9. Changes in FWHM and R with cycling of asmachined samples at a = 361 MPa (R = -1).
0.00 -0.05 -0.10 600

N (cycles)

FWHM - FWHM0

500 400 300

FWHM - FWHM0

400 300

FWHM - FWHM0 ()

-0.15 -0.20 -0.25 -0.30


R - R0
a = 319 MPa

FWHM - FWHM0 ()

-0.15 -0.20 -0.25 -0.30


a = 367 MPa, R = -2.8

R - R0 (MPa)

R - R0 (MPa)

200 100 0 -100 -200 -300 200000

200
R - R0

100 0

As-machined samples Run out


T1EF04 T1EF05 T1EF08 T1EF15 Average curve

-0.35 -0.40 -0.45 0 40000 80000 120000


T1EF16 run out T1EF17 run out T1EF18 run out

-0.35 -0.40 -0.45 0

As-machined samples T1EF06 T1EF07 T1EF14 Average curve

-100 -200 -300 70000

160000

10000

20000

30000

40000

50000

60000

N (cycles)

N (cycles)

Figure 7. Changes in FWHM and R with cycling of asmachined samples at a = 319 MPa (R = -1).
0.00 -0.05 -0.10 600

Figure 10. Changes in FWHM and R with cycling of asmachined samples at a = 367 MPa (R = -2.8).
0.05

FWHM - FWHM0

500

0.00
400

FWHM - FWHM0 ()

-0.20 -0.25 -0.30


a = 348 MPa

200 100 0 -100 -200 -300 70000

FWHM - FWHM0 ()

-0.15

300

-0.05

R - R0 (MPa)

-0.10
Average curves As-machined samples R = -1 a = 277 MPa
a = 319 MPa

R0 - R0

-0.15

-0.35 -0.40 -0.45 0

As-machined samples T1EF11 T1EF12 T1EF13 Average curve

-0.20
N1 N2

a = 348 MPa a = 361 MPa

10000

20000

30000

40000

50000

60000

-0.25 0 40000 80000 120000 160000 200000 240000 280000 320000

N (cycles)

N (cycles)

Figure 8. Changes in FWHM and R with cycling of asmachined samples at a = 348 MPa (R = -1).

Figure 11. Average changes in FWHM with cycling of asmachined samples at different stress amplitudes.

Average curves of changes in FWHM during fatigue tests at four different stress amplitudes (R = -1) are shown in Fig. 11. Increasing stress amplitude accentuates variations in FWHM and reduces the duration of each stage. The numbers of cycles at the end of Stages 1 and 2 are defined as N1 and N2, respectively. N1 and N2 for the four alternating stress loads with

R = -1, estimated from Fig. 11, are presented in Table 4, where Nf is the number of cycles at failure and FWHM1 FWHM0 is the variation in FWHM at N1. Fig. 12 shows that a linear correlation is found between N1 and N2 and the stress amplitude.

Copyright 2012 by ASME

Table 4. Numbers of cycles at the end of Stages 1 and 2 of FWHM changes during fatigue tests on as-machined samples (R = -1).

a
(MPa) 277 319 348 361
400

FWHM1- FWHM0 () -0.0377 -0.0487 -0.1078 -0.1259

N1 (cycles) 86,667 42,500 20,000 9,000

N2 (cycles) 230,000 122,500 56,000 45,000

Nf (cycles) 297,033 147,050 59,467 55,850

a x N1

Alternating stress amplitude (MPa)

380 360 340 320 300 280 260 0 40000 80000 120000 160000

As-machined samples Annealed samples


a x N2
As-machined samples

Annealed samples

200000

240000

280000

N1, N2 (cycles)

Figure 12. Numbers of cycles at the end of Stage 1 (N1) and Stage 2 (N2) during fatigue tests at different stress amplitudes.

Considering that microstructural changes measured in terms of variations in FWHM present three regular successive stages (Fig. 11), it can be supposed that a correlation could exist between them and the three phases of fatigue damage mechanisms, namely, initiation of microcracks, microcracking (microcrack propagation), and macrocrack propagation [38]. In the first phase, strain localization leads to the formation of persistent slip bands (PSBs), extrusions and intrusions, from which microcracking takes place. The second phase is characterized by microcracking and microcrack propagation (on the order of nm/cycle), usually along PSBs or slip planes of the crystalline structure. In the last phase, macrocracking and macrocrack propagation take place. Macrocrack propagation (on the order of m/cycle) occurs in a direction normal to the maximum tensile stress and leads to failure. In Stage 1 of the microstructural changes observed, the decrease in FWHM can be associated with the movement and multiplication of dislocations, rearrangement of the initial dislocation network, and reduction in microdeformations. The initial dislocation network would be produced by the pipe manufacturing process, which induces some level of cold working to the material. Microstructural changes during the early stages of fatigue are connected to the movement and reorganization of dislocations and to strain localization [9,39]. Lattice defects are one of the main mechanisms that result in FWHM changes [40]. The level of microdeformations can be indirectly characterized by X-ray diffraction peak broadening; a decrease in FWHM is related to a reduction in microdeformations [8]. From a macroscopic point of view, these changes are characterized by cyclic hardening or

softening, depending on the initial dislocation network state of the material [9]. An increase in FWHM is related to an increase in the work hardening level, produced by development of microdeformations mainly built up by increased dislocation density [41,42]. Cyclic hardening during fatigue tests was associated with increase in FWHM during cyclic deformation of under-aged AA6110 aluminum alloy [43]. At the same time, fatigue cycling can involve the decay in the dislocation density or cold work, characterized by a reduction in FWHM [40,44]. Work softening in shot peened hardened steels was found to be followed by a decrease in FWHM caused by rearrangement of high density dislocations into lower energy structures [40]. The evolution of FWHM was found to decrease in regular successive stages with fatigue cycling of cold-worked (hardened), shot peened, deep rolled, and laser shot peened steel samples [18,27,41,42,45-50]. In contrast, FWHM may increase in the early stage of fatigue for annealed materials, since cyclic loading will most likely increase lattice distortion, which is associated with an increase in the dislocation density. During Stage 2 the rate of decrease in FWHM is considerably reduced. This stage corresponds to the major fraction of fatigue life (about 50%) and can be associated with microcracking and microcrack propagation. This stage can be regarded as the propagation of a virtual crack, assuming that this crack would be the sum of individual microcracks propagating at a very low rate (on the order of nm/cycle). Supposing that the virtual microcrack size reaches about one or two grain sizes (10-20 m) at the end of Stage 2, it is possible to roughly estimate an apparent microcrack propagation rate during this stage. From data given in Fig. 11 and Table 4, propagation rates of about 0.07 and 0.30 nm/cycle were estimated for the highest and lowest stress amplitude, respectively. This represents a very interesting perspective, since it would be possible to predict the duration of Stage 2 and quantify fatigue damage before macroscopic cracking, which takes place during Stage 3. In the beginning of this stage microcracks coalesce to form a macrocrack that propagate up to failure. In this stage, the decrease in FWHM can be attributed to relaxation of microstresses due to macrocrack initiation and propagation preceding final failure. Figs. 6 to 10 show that the evolution of macro residual stresses (R) with fatigue cycling also presents three stages, with nearly the same durations of those observed in FWHM changes for a given stress amplitude. It can be considered that there is a direct correlation between changes in FWHM and residual stresses for all stress amplitudes. The evolution of residual stresses in Stage 1 could be different depending on the level of stress amplitude applied. For low stress amplitudes (277 MPa), residual stresses tend to decrease, while for high stress amplitudes (319-367 MPa) residual stresses tend to increase. Finally, in Stage 3, macroscopic cracking is followed by the release of residual stresses. Generally, when residual stresses (top surface) increased, the crack propagated in the sample bottom surface, whereas the residual stresses decreased when the crack propagated in the sample top surface. This is in agreement with the consideration that crack propagation would be accompanied by residual stress release along the crack.

Copyright 2012 by ASME

0.05
Average curves / 2 = 0.19% As-machined samples a = 361 MPa, R = -1
a = 367 MPa, R = -2.8

0.15

200 150 100 50


R - R0

0.00

0
a = 319 MPa

FWHM - FWHM0 ()

FWHM - FWHM0 ()

-0.05

0.10

-50

R - R0 (MPa)

-0.10

Annealed samples T1EF19 T1EF25 Average curve

-100 -150 -200 -250 -300 -350 -400 -450

-0.15

0.05

-0.20

FWHM - FWHM0
-0.25 0 10000 20000 30000 40000 50000 60000 70000

-500 80000 120000 160000 -550 200000

0.00 0 40000

N (cycles)

N (cycles)

Figure 13. Changes in FWHM with cycling of as-machined samples at strain amplitude of 0.19% under R = -1 and - 2.8. Table 5. Fatigue test results of annealed samples.

Figure 14. Changes in FWHM and R with cycling of annealed samples at a = 319 MPa (R = -1).
0.15 200 150 100 50
R - R0

Strain Sample amplitude Applied Corrected 1 T1EF19 0.17% T1EF25 T1EF26 0.19% T1EF27
1

Stress amplitude (MPa) R

N (cycles)
FWHM - FWHM0 ()

0 -50

319 361

319 0.61o -1.1


1

361 0.69o -1
1

173,500 68,500 34,100 26,500

0.10

R - R0 (MPa)

-100 -150 -200 -250 0.05 -300 -350 -400 -450 -500

Gerber mean stress criterion

FWHM - FWHM0

a = 361 MPa

In Fig. 13, average curves of changes in FWHM during fatigue tests obtained at the same strain amplitude (0.19%) with two different stress ratios R are compared. Even if the duration of each stage seems to remain unchanged, the decrease of the stress ratio (from -1 to -2.8) considerably reduced the amplitude of FWHM variations. It is considered that this behavior can be due to the lower tensile maximum stress at R = -2.8, which would result in less microstructural changes during fatigue cycling and, consequently, lower FWHM variations. Annealed samples Four alternating bending fatigue tests were carried out on annealed samples at two different stress amplitudes, a = 319 and 361 MPa (R = -1). Two tests were conducted for each load level. Table 5 presents the results of fatigue tests on annealed samples. Fatigue lives of annealed samples were lower than those of as-machined samples for the same stress amplitude due to annealing effects, which is known to result in a reduction of fatigue life and strength [51]. The evolution of FWHM and residual stresses with fatigue cycling on annealed samples at stress amplitudes of 319 and 361 MPa (R = -1) is shown in Figs. 14 and 15, respectively. Similarly to results obtained for as-machined samples, the evolution of FWHM with fatigue cycling on annealed samples also presents three stages. However, instead of a decrease in FWHM in the first and third stages, as observed for asmachined samples, an increase in FWHM is observed for annealed samples. In the second stage, a decrease in FWHM is observed for both as-machined and annealed samples.

Annealed samples T1EF19 T1EF25 Average curve

0.00 0 5000 10000 15000 20000 25000 30000

-550 35000

N (cycles)

Figure 15. Changes in FWHM and R with cycling of annealed samples at a = 361 MPa (R = -1).

The transformation of the material to a lower energy state and the restoration of the dislocation network state by annealing reduced the work hardening effect generated by the pipe manufacturing process, as indicated by lower average values of FWHM of annealed samples (1.83) with respect to as-machined samples (2.06). As a result, the first stage of the evolution of FWHM for annealed samples is characterized by an increase in FWHM, as fatigue loading enhances the energy state of the material. This stage (Stage 1) could be related to multiplication of dislocations, intensification of their interactions and hardening of the material, which result in a fast increase in FWHM. Stage 1 could be associated with the development of PSBs, extrusions and intrusions. In the second stage (Stage 2), the decrease in FWHM can be associated with the rearrangement of the dislocation network, with preponderant annihilation of dislocations, and microcracking (microcrack propagation). Finally, FWHM values increased in the third stage (Stage 3) until complete fracture, indicating that the creation of new free surfaces during macrocrack propagation follows the rising energy of the material. The rate of changes in FWHM is accentuated with increasing stress amplitude. Figure 12 shows that a linear correlation between N1 and N2 and the stress amplitude is also found for annealed samples.

Copyright 2012 by ASME

Figure 16. TEM lamella produced from sample T1EF18.

microscopy (TEM) analyses combined with focused ion beam (FIB) technique to investigate the correlation between the evolution of FWHM values and microstructural changes (dislocation density and structure) with cycling. Samples were cycled at a stress amplitude of 319 MPa (R = -1) to 20,000 (T1EF16), 80,000 (T1EF17), and 120,000 cycles (T1EF18). A FEI Dual Beam (FIB/SEM) system, combining focused ion beam with scanning electron microscopy techniques, was used to produce thin electron transparent lamellae to be analyzed by means of TEM. TEM lamellae were prepared from the center of the sample gage length. Lamella preparation involved milling the surface with gallium ions to obtain the lamella section, extraction of the lamella with the aid of a mechanical needle, thinning, and sticking the electron transparent lamella 80-100 nm thick to the TEM specimen-holder. In Fig. 16 the lamella produced from sample T1EF18 is shown in the TEM specimenholder. Figures 17 and 18 present TEM images of lamellae produced from samples T1EF16 and T1EF18, respectively. It is possible to identify the fatigue damage effect, observing the high density dislocations and more entangled dislocation structure in sample T1EF18 (120,000 cycles) with respect to sample T1EF16 (20,000 cycles). A qualitative evaluation of dislocation structures suggests that the dislocation density increased between 20,000 and 120,000 cycles. CONCLUSIONS The aim of the work is to evaluate and quantify the evolution of microdeformations and macro residual stresses during high cycle fatigue (HCF) life of API 5L X60 grade steel samples using the X-ray diffraction (XRD) technique. Samples are submitted to strain-controlled alternating bending fatigue tests at room temperature. Fatigue test samples are prepared under two different conditions: as-machined and annealed. Microdeformations are evaluated from measurements of the full width at half maximum (FWHM) of the XRD peak and macro residual stresses are estimated according to the sin method. Considering that microstructural changes measured in terms of variations in FWHM present three regular successive stages, for both as-machined and annealed samples, a correlation could be supposed between them and the three phases of fatigue damage mechanisms, namely initiation of microcracks, microcracking (microcrack propagation), and macrocrack propagation. Stage 1 can be associated with the movement and multiplication of dislocations and rearrangement of the initial dislocation network. Stage 2 corresponds to about 50% of fatigue life and is characterized by a smooth decrease in FWHM. This stage, related to microcracking and microcrack propagation, can be considered as the propagation of a virtual crack, assuming that this crack would be the sum of individual microcracks propagating at a very low rate (on the order of nm/cycle). Assuming that the virtual microcrack size reaches about one or two grain sizes (10-20 m) at the end of Stage 2, it is possible to roughly estimate an apparent microcrack propagation rate during this stage. This represents a very interesting perspective, since it would be possible to predict the duration of Stage 2 and quantify the fatigue damage before macroscopic cracking (Stage 3). After Stage 2, microcracks coalesce to form a macrocrack, which propagates during Stage

Figure 17. Dislocation structure in sample T1EF16 cycled at a = 319 MPa (R = - 1) to 20,000 cycles.

Figure 18. Dislocation structure in sample T1EF18 cycled at a = 319 MPa (R = - 1) to 120,000 cycles.

TEM ANALYSES Three as-machined samples were fatigued to specific numbers of cycles, representing the three stages of microstructural changes (Fig.7), for transmission electron

Copyright 2012 by ASME

3. Microstructural changes show to be well correlated to the level of stress amplitude applied. From TEM analyses of fatigued test samples, it is possible to qualitatively evaluate the fatigue damage effect, considering the high density dislocations and more entangled dislocation structure observed with increasing number of cycles. Care should be taken to correlate X-ray diffraction data to dislocation density and structure under fatigue loading conditions. To date, this is only established under monotonic loading conditions [23-25]. Up to now, FWHM and residual stresses measured during fatigue cycling should be discussed solely in the light of damage evolution. The objective of this ongoing work is the quantification of variations in FWHM and residual stresses, as well as their durations, measured during HCF cycling and the correlation between them and the loading level (stress amplitude). At the same time, a correlation between FWHM and the dislocation density and structure is supposed to exist also in fatigue loading conditions. To verify this correlation, quantitative TEM investigations are envisaged in a second phase of this ongoing work. The results obtained show that the X-ray diffraction technique, even if it cannot elucidate by itself the whole nature of the phenomenon, can be used as a tool to evaluate fatigue damage in real time. This was rendered possible by the portability and speed of the X-ray diffraction equipment used. X-ray diffraction could be a useful means of bridging the seemingly dissociated macroscopic and microscopic approaches to fatigue studies that have been usually carried out independently until now. These results offer interesting perspectives for the future work in the objective of quantifying microstructural changes during fatigue life of steel structures or mechanical components. ACKNOWLEDGMENTS The authors would like to thank CNPq (Brazilian National Council for the Scientific and Technological Development) and Roberto Rocca Education Program (Tenaris Company) for the financial support, and the technical staff of the Interdepartmental Laboratory of Electron Microscopy (LIME), University of Rome "ROMA TRE", for the TEM analyses. NOMENCLATURE lattice spacing in nonstressed material d0 N number of load cycles at failure number of cycles at the end of Stage 1 N1 number of cycles at the end of Stage 2 N2 number of cycles at failure Nf R stress ratio, min / max alternating stress that causes failure at N cycles Sn ultimate tensile strength Su a alternating stress amplitude yield strength REFERENCES [1] D. Lyons, Western European Cross-country Oil Pipelines 30-year Performance Statistics, Report No. 1/02, CONCAWE, 2002.

[2] K. Paulson, A Comparative Analysis of Pipeline Performance, 2000-2003, National Energy Board, Alberta, Canada, 2005. [3] S.B. Cunha, I.P. Pasqualino, B.C. Pinheiro, Stress-life fatigue assessment of pipelines with plain dents, Fatigue Fract. Eng. Mater. Struct. 32 (2009) 961-974. [4] B.C. Pinheiro, I.P. Pasqualino, Fatigue analysis of damaged steel pipelines under cyclic internal pressure, Int. J. Fatigue 31 (2009) 962-973. [5] American Petroleum Institute, API Specification 5L, 42nd ed., 2004. [6] P.C. Paris, M.P. Gomez, W.E. Anderson, A rational analytic theory of fatigue. Trend Eng. 13 (1961) 9-14. [7] DNV, Fatigue Design of Offshore Steel Structures, Recommended Practice DNV-RP-C203, Hvik, Norway, 2008. [8] N. Ji, J.L. Lebrun, Microstructural study of static and dynamic deformed polycrystalline copper by X-ray diffraction profile analysis, Scr. Metall. Mater. 24(8) (1990) 1547-1552. [9] J. Lesage, D. Chicot, P. Perez, J.M. Chatelet, X-ray diffraction study of microstructural modifications during fatigue. In: Proceedings of 11th Conference on Materials Testing, v. 2 pp. 616-620, Balatonszplak, Hungary, 1994. [10] T. Magnin, J. Driver, J. Lepinoux, L.P. Kubin, Aspects microstructuraux de la dformation cyclique dans les mtaux et alliages C.C. et C.F.C. I. Consolidation Cyclique, Rev. Phys. Appl. 19 (1984) 467-482. [11] P.K. Liaw, H. Wang, L. Jiang, et al., Thermographic detection of fatigue damage of pressure vessel at 1,000 Hz and 20 Hz, Scr. Mater. 42 (2000) 389-395. [12] S. Luxemburger, W. Arnold, Laser ultrasonic absorption measurement in fatigue-damaged materials, Ultrasonics 40 (2002) 797-801. [13] H. Ogi, T. Hamaguchi, M. Hirao, Ultrasonic attenuation peak in steel and aluminum alloy during rotating bending fatigue, Metall. Mat. Trans. A 31 (2000) 1121-1128. [14] A. Gilanyi, K. Morishita, T. Sukegawa, et al., Magnetic nondestructive evaluation of fatigue damage of ferromagnetic steels for nuclear fusion energy systems, Fusion Eng. Des. 42 (1998) 485-491. [15] E.S. Palma, T.R. Mansur, S.F. Silva Jr, A. Alvarenga Jr, Fatigue damage assessment in AISI 8620 steel using Barkhausen noise, Int. J. Fatigue 27 (2005) 659-665. [16] A. Benrabah, C. Langlade, A.B. Vannes, Residual stresses and fretting fatigue, Wear 224 (1999) 267-273. [17] Y. Isobe, A. Kamimura, K. Furuta, E. Ikuta, K. Aoki, F. Nakayasu, Fatigue damage characterization in alloy 718, Superalloys 718, 625, 706 and various derivatives, ed. E.A. Loria, (1994) 607-618. [18] B. Scholtes, Residual stress analysis - a useful tool to assess the fatigue behavior of structural components, Adv. X-ray Anal. 43. In: Proceedings of the Denver X-ray Conference, Colorado, USA, 2000. [19] S. Rai, B.K. Choudhary, T. Jayakumar, K. Rao, B. Raj, Characterization of low cycle fatigue damage in 9Cr1Mo ferritic steel using X-ray diffraction technique, Int. J. Press. Vessels Pip. 76 (1999) 275-281.

Copyright 2012 by ASME

[20] K. Vijayan, A. Mani, C. Balasingh, A.K. Singh, X-ray analysis of polycrystalline aluminium subjected to fatigue cycling, Bull. Mater. Sci. 10(3) (1988) 205-216. [21] G. Maeder, J.L. Lebrun, J.M. Sprauel, Present possibilities for the X-ray diffraction method of stress measurement, NDT Int. 14(5) (1981) 235-247. [22] N. Broll, Caractrisation de Solides Cristalliss par Diffraction X, PE 1080, Techniques de lIngnieur, Trait Analyse et Caractrisation, 1996. [23] T. Ungr, Dislocation densities, arrangements and character from X-ray diffraction experiments, Mat. Sci. Eng., A 309-310 (2001) 14-22. [24] B. Hoffmann, 0. Vijhringer, E. Macherauch, Effect of tempering on the microstructure and strength of martensitically hardened plain carbon steels, Mater. Sci. Eng. A234-236 (1997) 707-710. [25] M.B. Kerber, E. Schafler, M. Zehetbauer, Processing and Evaluation of X-ray Line Profiles Measured from Nanostructured Materials Produced by Severe Plastic Deformation, Rev. Adv. Mater. Sci. 10 (2005) 427-433. [26] D.J. Quesnel, M. Meshii, J.B. Cohen, Residual stress in high strength low alloy steel during low cycle fatigue, Mater. Sci. Eng. 36 (1978) 207-215. [27] I. Nikitin, I. Altenberger, Comparison of the fatigue behavior and residual stress stability of laser-shock peened and deep rolled austenitic stainless steel AISI 304 in the temperature range 25-600C, Mat. Sci. Eng., A 465 (2007) 176-182. [28] M. Krauss, B. Scholtes, Thermal shock damage of hotwork tool steel AISI H11 in hard turned, electroeroded, shot-peened or deep rolled surface conditions, J. Mater. Sci. Technol. 20(1) (2004) 93-96. [29] B. Pinheiro, J. Lesage, I. Pasqualino, N. Benseddiq, Assessment of Fatigue Damage Initiation in Oil and Gas Steel Pipes, OMAE2011-49044, In: Proceedings of the 30th International Conference on Ocean, Offshore and Arctic Engineering, June 19-24, 2011, Rotterdam, The Netherlands. [30] B. Pinheiro, J. Lesage, I. Pasqualino, N. Benseddiq, E. Bemporad, X-ray diffraction study of microstructural changes during fatigue damage initiation in steel pipes, Mat. Sci. Eng., A 532 (2012) 158-166. [31] D. Tabor, The hardness and strength of metals, J. Inst. Met. 79 (1951) 1-18. [32] J.R. Cahoon, An improved equation relating hardness to ultimate strength, Metall. Trans. 3 (1972) 3040. [33] J. Moteff, R.K. Bhargava, W.L. McCullough, Correlation of the hot-hardness with the tensile strength of 304 stainless steel to temperatures of 1200C, Metall. Trans. A 6A (1975) 1101-1104. [34] C. Schenck, Wechselbiegemaschine PWON, Maschinenfabrik GmbH, Darmstadt, Germany, 1971. [35] AFNORAssociation Franaise de Normalisation, Essais non destructifs Mthodes dessais pour lanalyse des contraintes rsiduelles par diffraction des rayons X, XP A 09-285, 1999. [36] J.E. Shigley, C.R. Mischke, Mechanical Engineering Design, 6th ed., McGraw-Hill, New York, 2001.

[37] C.S. Vianna, Mechanical Behavior of the API 5L X-60 Steel with and without Hydrogen, M.Sc. Thesis, COPPE/UFRJ, Rio de Janeiro, Brazil, 2005. [38] G.E. Dieter, Mechanical Metallurgy, SI Metric ed., McGraw-Hill, London, 1988. [39] D. Chicot, J. Lesage, Diffraction X et modifications microstructurales en cours de fatigue, X-ray diffraction and microstructural modifications during fatigue, In proceedings: Innovation Technologique pour les Transports Terrestres, Technological Innovation for Land Transportation (TILT 2003), 2-4 dcembre 2003, Lille, France, 803-810. [40] D. Lhe, O. Vhringer, Stability of Residual Stresses, In: ASM International Handbook of Residual stress and Deformation of Steel, G. Totten, M. Howes, T. Inoue (ed.), 54-69, ASM International, USA, 2002. [41] K. Dalaei, B. Karlsson, L.-E. Svensson, Stability of residual stresses created by shot peening of pearlitic steel and their influence on fatigue behavior, Procedia Eng. 2 (2010) 613-622. [42] K. Dalaei, B. Karlsson, L.-E. Svensson, Stability of shot peening induced residual stresses and their influence on fatigue lifetime, Mater. Sci. Eng. A 528 (2011) 1008-1015. [43] P. Juijerm, I. Altenberger, Cyclic Deformation Behaviour and Its Optimization at Elevated Temperature, In: Aluminium Alloys, Theory and Applications, Tibor Kvackaj (ed.), InTech, 183-198, 2011. [44] R. Lemaitre, J.-L. Lebrun, J. Maeder, Contraintes Rsiduelles et Fatigue, Matriaux et Techniques, 297-304, 1982. [45] H. Holzapfel, V. Schulze, O. Vhringer ,E. Macherauch, Residual stress relaxation in an AISI 4140 steel due to quasistatic and cyclic loading at higher temperatures, Mater. Sci. Eng. A 248 (1998) 9-18. [46] A. Wick, V. Schulze, O. Vhringer, Effects of warm peening on fatigue life and relaxation behaviour of residual stresses in AISI 4140 steel, Mater. Sci. Eng. A 293 (2000) 191-197. [47] A. Cherif, Y. Pyoun, B. Scholtes, Effects of ultrasonic nanocrystal surface modification (UNSM) on residual stress state and fatigue strength of AISI 304, J. Mater. Eng. Perform. 19(2) (2010) 282-286. [48] I. Nikitin, M. Besel, Correlation between residual stress and plastic strain amplitude during low cycle fatigue of mechanically surface treated austenitic stainless steel AISI 304 and ferriticpearlitic steel SAE 1045, Mater. Sci. Eng. A 491 (2008) 297-303. [49] I. Nikitin, M. Besel, Residual stress relaxation of deep rolled austenitic steel, Scr. Mater. 58 (2008) 239-242. [50] I. Nikitin, B. Scholtes, H.J. Maier, I. Altenberger, High temperature fatigue behavior and residual stress stability of laser-shock peened and deep rolled austenitic steel AISI 304, Scr. Mater. 50 (2004) 1345-1350. [51] D. Lhe, O. Vhringer, Residual Stresses and Fatigue Behavior, in: ASM International Handbook of Residual stress and Deformation of Steel, eds. G. Totten, M. Howes, T. Inoue, ASM International, Materials Park, Ohio, 2002, pp. 27-53.

10

Copyright 2012 by ASME