Fatigue of X65 steel in the sour corrosive environment—A novel experimentation and analysis method for predicting fatigue crack initiation life from corrosion pits

Northumbria University has developed Northumbria Research Link (NRL) to enable users to access the University’s research output. Copyright © and moral rights for items on NRL are retained by the individual author(s) and/or other copyright owners. Single copies of full items can be reproduced, displayed or performed, and given to third parties in any format or medium for personal research or study, educational, or not-for-profit purposes without prior permission or charge, provided the authors, title and full bibliographic details are given, as well as a hyperlink and/or URL to the original metadata page. The content must not be changed in any way. Full items must not be sold commercially in any format or medium without formal permission of the copyright holder. The full policy is available online: http://nrl.northumbria.ac.uk/pol  i cies.html


| INTRODUCTION
API-5L Grade X65 steel are commonly used to manufacture pipelines used to convey production fluids extracted by the oil and gas industry. 1,2 The mechanical properties, cost, and availability are the main factors that determine the selection of steel. However, despite having modest resistance to uniform corrosion, this grade of steel is prone to localized corrosion (pitting) upon exposure to sour corrosive environments, 3,4 typical for many mature oil and gas reservoirs. The sour corrosive media can reduce the fatigue life of pipelines under the cyclic load 5 applied from sea waves, seabed movements, and internal pressure variation. 6 In order to increase operational safety and minimize the likelihood of pipelines failures, a better informed and more efficient maintenance and inspection schedule is required. A vital prerequisite on the path to meet these goals is the development of a more reliable fatigue life prediction model.
The presence of corrosion pits has been cited as main contributory factor in many failures of pipelines reported by the oil and gas industry. 2,3 Corrosion pits are a form of localized corrosion that can jeopardize the life of engineering assets by accelerating material loss over a small area or by increasing the risk of crack initiation as a stress riser. A major part of the total fatigue life comprises the transition of corrosion pit to fatigue cracking, 7 the so-called pitto-fatigue crack transition that has received growing attention in recent research publications. 8 Different models exist in the literature to predict crack initiation life from corrosion pit, [9][10][11][12][13][14] which are based on linear elastic fracture mechanics (LEFM) criteria. This criterion predicts the behavior of the long crack by considering corrosion pit as a pre-existing short crack and disregards the crack initiation regime which is quite a significant regime in high cycle fatigue life. The model proposed in this paper for predicting the crack initiation life from corrosion pit in a sour environment considers pits as a stress concentration zone, that is, a notch, and predicts short crack initiation life. This model uses the data obtained from environmental fatigue testing that replicates sour conditions present in service.
Previous studies have described fatigue tests performed in air or environmental fatigue tests conducted in benign environments 10,[14][15][16][17] investigating the effect of the presence of corrosion pits. In earlier publications 18,19 we have summarized the discrete parts of our initial efforts toward developing a harmonized approach to environmental fatigue testing of alloys. Prior to our work, there were no reports of small-scale corrosion fatigue tests of alloys exposed to toxic H 2 S containing fluids, because of the test complexity and health and safety considerations. In this paper, we detail the overall approach for the first time and present new data concerning the effect of sour environments on fatigue behavior of materials using small-scale standard samples.

| Material and specimens
The material of interest in this study is seamless API-5L X65 grade pipeline steel provided by the industrial sponsor. The chemical composition was obtained by optical emission spectroscopy at The Welding Institute (TWI Ltd.) reported in previous work 20 and the cyclic and monotonic mechanical properties required in this paper were obtained from the literature 7 as reported in Table 1. The chemical composition of tested material is reported in Table 2.
All experiments were carried out on specimens extracted longitudinally from parent material of an X65 steel pipe ( Figure 1A,B). The seamless pipe had an outer diameter of 273 mm and wall thickness of 29 mm. Only parent material remote from the girth weld was used in the present study. Two types of fatigue specimens were used during environmental tests, namely, smooth ( Figure 1C) and pre-pitted ( Figure 1D); both were designed in accordance with ASTM E466 standard. To make a smooth contact between the test apparatus o-rings and the specimen, a curved edge was used on all specimens. Initial trials indicated that, in order to decrease the number of artifacts in X-ray computed tomography images and to optimize the resolution, it was necessary to reduce the width of the pre-pitted specimens to 3 mm ( Figure 1D). All test specimens were ground to a 4000 paper finish (carbon-silica) prior to a final polish using a 3-μm cloth to minimize the potential effects of residual stress and surface roughness on fatigue behavior.

| Pre-pitting of the specimens using an electrochemical cell
A number of specimens used in fatigue tests were prepitted using a micro-electrochemical cell constructed at TWI Ltd. The electrolyte used during these studies was 3.5% NaCl solution (exposed to ambient air). Figure 2 shows the main components of the electrochemical workstation including the electrochemical microcell and Figure 3 is a close-up view of the microcell. The cell was installed on a VersaSCAN device that allows precise positioning of the cell along three orthogonal axes. A standard three-electrode configuration (i.e., working, counter and reference electrodes) was employed, using a platinum wire as the counter electrode and an Ag/AgCl reference electrode. All the electrodes are attached to Ver-saSTAT 4 potentiostat-galvanostat. Both the VersaSCAN and VersaSTAT 4 were controlled using VersaStudio software and a laptop PC. The purpose of the microcell is to control the size of the wetted area of the specimen surface, by using a micro-capillary to form a droplet that can be positioned on the specimen surface and thus impart region-selectivity to the electrochemical corrosion process. A video camera accessory is used to accurately position and monitor the droplet at the specimen surface prior to and during the electrochemical corrosion (i.e., pitting) process. Figure 4 shows a close-up view of a pre-pitted specimen. Following polarization, the specimens were placed in a Pyrex glass beaker containing acetone and cleaned by ultrasonication. The as-grown pits were characterized using an Alicona Laser Scanning Confocal Microscope with resolution of 0.1 micron. The pit depth and lateral dimensions were recorded across four directions as shown in Figure 5A. The average of the measured values for the four profiles was used to define the final depth and width of pits. A representative confocal microscopy scan and a 2D-cross-sectional profile are shown in Figure 5B,C, respectively. The pits generated using this method are morphologically similar to the pits seen on the internal surface of the pipes; in terms of pit depth, width and cross-sectional shapes of corrosion pits as observed in samples and data (i.e., data collected using non-destructive testing (NDT) methods on steel production risers) provided by the industrial sponsor of this project.

| In situ corrosion fatigue test apparatus
A bespoke environmental apparatus was developed to facilitate the environmental tests and visualize the initiation of a fatigue crack from a corrosion pit in X65 steel using X-ray CT. The development of the apparatus including material and size selection is detailed in previous work. 18 The provision of robust sealing for the test cell was a vital consideration, due to the toxic nature of the H 2 S contained therein and the need to exclude oxygen from the test environment. The configuration employed in this work ensured that the contents of the cell were fully isolated from the external environment throughout testing using axially loaded small-scale tensile specimens. The overall height of the vessel was chosen to be 130 mm to accommodate the specimen dimensional requirements stated in ASTM E466. The specimen's dimensions are stated in Section 2.1 and Figure 1. Figure 6 shows the vessel placed inside the fatigue test machine. The gas/ solution inlets and outlets are also shown.

| Fatigue test procedure in the sour environment
A corrosion pit of the desired dimensions was created in the center of the test section of the specimen by using the electrochemical cell (see Section 2.2). Table 3 shows the size of corrosion pits of the 10 specimens for corrosion fatigue test. An initial corrosion fatigue test was performed using the smooth specimen to both check the performance of the apparatus and also to obtain an estimate of the number of cycles to failure in a corrosive environment. The specimens were tested under the constant amplitude loading condition and sinusoidal waveform in the high cycle fatigue regime. The applied stress ratio (R) was 0.1 for all tests. The applied stress amplitude ranged from 135 to 225 MPa (corresponding maximum applied stress from 300 to 500 MPa) for the 10 smooth specimens with cross-sectional area of 3 × 10 mm 2 ( Figure 1C). The 10 pre-pitted specimens with cross-sectional area 3 × 3 mm 2 ( Figure 1D) were tested at two different applied stress amplitudes of 165 and 185 MPa, corresponding to maximum applied stress 367 MPa (70% YS) and 412 MPa (80% YS), respectively, at the gauge section. The yield strength is 516 MPa. Table 3 shows the applied stress amplitudes for the pre-pitted samples. A frequency of 0.3 Hz was applied in order to allow the corrosive environment to interact with the specimen. Testing was conducted at ambient pressure and temperature. The test environment comprised 3.5% w/v NaCl solution saturated with a gas mixture containing 12.5% H 2 S in CO 2 balance.
The sour corrosion fatigue tests were performed in the Gooch laboratory at TWI Ltd. in Cambridge, which has fixed H 2 S sensors for monitoring and suitable extraction and safety procedures in place. Prior to starting the test, 3.5% NaCl solution was deaerated by purging with high-purity nitrogen for 24 h in a sealed High-Density Polyethylene (HDPE) barrel. Previous work at TWI Ltd. has shown that this process reduces the measured dissolved oxygen content to below 10 ppb. 20 For each test, once the apparatus and specimen were assembled and placed in the fatigue loading machine, a leak test was carried out by pressurizing the test cell with nitrogen to 0.5 barg to ensure that the seals did not leak. Following confirmation of seal integrity, the test cell was deaerated using a fast purge of nitrogen for 1 h, then charged with 3.5% NaCl solution, and then purged once again with nitrogen for approximately 1 h. Upon completion of this process, the test solution in the cell was saturated with sour test gas (i.e., containing 12.5% H 2 S) using a fast purge for 1 h. Finally, the fatigue load was applied to the specimen. The vessel contains about 380 cm 3 of the solution. The experimental setup is reported in previous work. 18

| Environmental fatigue testing on smooth and pre-pitted samples
Using the test protocol described in Section 2.4, in situ corrosion fatigue tests were carried out on 10 smooth  specimens to obtain S-N data in the sour environment.
The results are plotted in Figure 7. The fatigue test carried out in air on X65 steel in authors' previous work is described in full in previous work. 21 It can be seen that in the sour environment, a significant reduction in fatigue strength is apparent for all stress levels investigated in comparison to the corresponding fatigue strength tested in the air. The best fitted trend line for experimental data with regression of R 2 = 0.78 was obtained using a power law type equation that is displayed on the graph. Other types of functions showed R 2 being 0.62 for exponential, 0.59 for linear, and 0.71 for polynomial relations. Interestingly, comparison of the best fitted sour environment S-N curve with the one obtained in the air condition indicates that the fatigue strength reduction was more significant at higher fatigue lives. This finding is consistent with literature studies in this area where a corrosive environment was shown to reduce fatigue life more significantly at lower stress amplitudes. 22,23 This happens because at lower stress amplitudes, the environment has more time to interact with the material and in the case of sour environments, ingress of embrittling atomic hydrogen is anticipated. The Basquin equation for sour environment S-N curve that is obtained from the best fitted line with the test data is where σ a is the applied stress amplitude and N f is the number of cycles to failure, at R = 0.1. This empirical equation calibrated on this first set of results was used for the life prediction of pre-pitted specimens reported in the later section. Figure 7 also shows the environmental fatigue tests on pre-pitted specimens described in Table 3 at two different stress amplitudes, 165 MPa and 185 MPa. The results are compared with the smooth specimens tested in the same sour environment as well as in the air. This figure shows lower fatigue strength of pre-pitted samples compared to the smooth samples tested in sour environment that emphasizes the effect of corrosion pits in lowering the fatigue strength as a stress raiser.
Previous work by the authors has shown the initiation of fatigue crack from corrosion pits in pre-pitted samples, captured by X-ray tomography by interrupting the environmental fatigue tests. 18 It was observed that crack initiation life, N i , was about 90% of the total number of cycles to final failure, N f . Therefore, on consideration of this observation and also the small thickness of the fatigue test specimens, N i in this paper is approximated as 90% N f .
To study the effects of applied stress, pit geometry and pit size on fatigue crack initiation life, Murakami's square-root-area parameter model was used in F I G U R E 7 Stress versus number of load cycles to failure for X65 steel tested at R = 0.1 in the air and sour environment (smooth and pre-pitted specimens) [Colour figure can be viewed at wileyonlinelibrary.com] calculating the stress intensity factor range, that is, using Equation 2 24,25 : where ΔS is the applied stress range and c is half of the pit mouth width, that is, pit mouth radius. Figure 8 shows the relation between the stress intensity factor range and crack initiation life. A good correlation is established for all test data covering two different applied stress levels and a range of pit geometry and sizes as given in Table 3. This is because that the stress intensity factor takes account of both the applied cyclic stress and the pit size.
This result is rather significant in the context of fatigue data collected in sour corrosive environments and, in addition, obtained using small scale coupon samples for the first-time.

| Prediction model for crack initiation life from corrosion pit
The notch stress approach was used to predict the crack initiation life of the pre-pitted specimens tested in the sour corrosive environment. First, the local stress amplitude at corrosion pits was calculated by Glinka's formula (Equation 3) 26,27 where ΔS and Δσ are applied and local stress range, E the elastic modulus,ń the cyclic strain hardening exponent, andḰ the cyclic strength coefficient. The required material properties in these equations are reported in Table 1. The required stress concentration factor, K t , was calculated using Equation 4. 28 Good agreement was found between the notch stress approach and authors' finite element analysis. 19,21 In previous literature, 20 it is also shown that the pit local stress ratio obtained from Glinka's formula, at applied stress amplitude of 165 MPa and higher, is very close to the one obtained by FEA.
K t = 1 + 1:25 Thereafter, the obtained local stress amplitude was used to predict the fatigue crack initiation life (N i ) using the material's S-N data measured in the sour environment as presented in Figure 7 and Equation 1 (i.e., by substituting the obtained local stress amplitude in Equation 1, N f was calculated that is approximated as N i for prediction model). The function for the prediction line in Figure 9 is then equal to Equation 1. In this work the crack initiation stage is considered as the initiation of a short crack with length of the order of 1 mm. The predicted life is shown in Figure 9 in terms of the pit local stress amplitude calculated by the aforementioned notch stress method. The experimental results of pre-pitted samples are also shown in this F I G U R E 8 Stress intensity factor range versus number of cycles to crack initiation for pre-pitted samples tested in this study [Colour figure can be viewed at wileyonlinelibrary.com] figure in terms of local stress amplitude at the pits. It can be seen that the prediction is in a good agreement with the experimental test results. For a given life of experimental data, the relative error in local stress amplitude is between 7% and 22% with the mean difference of 16% and standard deviation of 13 MPa. The trend line in Figure 9 is obtained from best fitted line with experimental data and has a power type function with highest possible regression of R 2 = 0.67. Figure 10 shows number of cycles from experiments versus prediction. The envelope of data in this figure is on the conservative side, that is, right side of the y = x graph. This shows that proposed prediction model does not overestimate the crack initiation life in sour environment. The predicted life is shorter than test measured; that is, prediction is conservative. Overall, these results indicate the dependency of crack initiation life on the pit geometry and applied stress for X65 steel exposed to a sour corrosive environment. The comparison of the prediction with F I G U R E 9 Comparison of predicted crack initiation life with the experimental test in the sour environment (pre-pitted specimens); the prediction is made in terms of the pit local stress calculated by the notch stress method, equation 3 F I G U R E 1 0 Comparison of experiment with prediction in terms of the number of cycles to crack initiation experimental data provides good evidence that the proposed method for predicting the life of crack initiation from corrosion pit is a deterministic approach for X65 steel in the sour corrosive environment.

| Qualitative fractography
Following specimen failure, a fractographic analysis was undertaken using scanning electron microscopy (SEM). The purpose of fractography in this work was to understand the possible crack initiation region in the corrosion pit, that is, whether it is from the mouth of the pit or from the wall, and the predominant crack growth direction. Previous research on crack initiation life from corrosion pit of three NiCrMoV steel using X-ray microtomography confirmed that the crack initiated from the pit mouth and grew in the longitudinal direction. 29 If this is the case in X65 steel pipes in sour environment, it can cause removal of the material from the surface rather than through thickness crack growth and rupture of the pipes. Prior to microscopy, specimens were cleaned by ultrasonication in deionized water and a commercial reagent (Pyrene) to remove any surface contamination. Figure 11 provides an overview of the main surface features observed for fractured specimens. Features include the crack origin (nucleation site), ratchet marks, fatigue crack propagation, and fast fracture zones. Figure 11A shows that crack was initiated from the corner of smooth samples tested in sour environment. Corrosion pits or other stress raisers cannot be seen in the crack initiation region.
In all case, crack originated at the corrosion pit and subsequently propagated across the fatigue crack growth zone. The absence of crack growth marks (beach marks) in this zone demonstrates that there was no variation in the applied stress during tests. The plane of the fatigue crack growth zone was developed perpendicular to the plane of applied stress. Finally, cracks developed and grew to a point where brittle fracture can occur in the fast fracture zone resulting in rapid crack propagation. Another surface feature that is commonly found in fatigue failures is the ratchet marks, which represent the coalescence boundary between two adjacent cracks initiated on different geometric planes. An example of ratchet marks between two crack initiation sites is given in Figure 12A. 30 Figure 12B show the ratchet marks and crack origin for specimen tested at stress amplitude of 186 MPa. As can be seen, there are multiple crack nucleation sites around the pit. SEM images of other specimen fracture surfaces reveal the same pattern of damage with crack origins around the corrosion pits. The SEM images from the top view of the specimens (the view of the sample shown in Figure 4) show that other cracks are initiated independent of the primary crack at the corrosion pit in other geometric planes close to the center of the pit ( Figure 13). However, these features are not dominant cracks and did not contribute directly to the final failure.

| CONCLUSIONS
The objectives of this work were to assess the effect of a sour corrosive environment, representing typical inservice conditions of oil and gas pipes, on the fatigue crack initiation life from corrosion pit for X65 steel and to develop a predictive method for the crack initiation life. Environmental fatigue testing was carried out on small-scale coupon samples using a bespoke testing apparatus at two different stress amplitudes, 165 and 185 MPa, and 0.3 Hz frequency. The test environment comprised 3.5% w/v NaCl solution saturated with a gas mixture containing 12.5% H 2 S in CO 2 balance. This study has produced an environmental S-N curve for the material and proposed a validated predictive model based on the notch stress approach. Key findings are as follows: • A correlation was found between the crack initiation life and the pit in service conditions (i.e., pit geometry, pit mouth width, applied stress, and test environment). • The average difference between the predicted life and experimental test result was about 16% and the prediction was on the conservative side. • Fractography examination confirms that there were multiple crack nucleation sites around the pit. Secondary cracks also initiated independent of the primary crack at the corrosion pit in other geometric planes close to the center of the pit. However, these features were not dominant cracks and did not contribute directly to the final failure.
These findings have significant implications for the understanding of crack initiation life from corrosion pit. This work presents the first effort toward developing a harmonized approach to environmental fatigue testing of alloys, through an iterative procedure including modeling and small-scale testing. This approach will prove useful in expanding this testing protocol for other harsh environmental fatigue testing and the predictive model certainly can be used for other materials. Presumably, this method can be used for different alloys exposed to sour and other hydrogen charging conditions. There are also no obvious barriers to its application across other environmental fatigue systems where pitting is anticipated under similar cyclic loading state. It should be noted that the pre-pitting method proposed in this work should be developed according to each specific application to ensure achieving similar pit shape and sizes as in-service condition for each application.

ACKNOWLEDGEMENT
The first author would like to thank BP, TWI Ltd., and Coventry University for the sponsorship of this project. The work was enabled through, and undertaken at, the National Structural Integrity Research Centre (NSIRC), a postgraduate engineering facility for industry-led research into structural integrity established and managed by TWI Ltd. through a network of both national and international Universities. We also acknowledge Dr. Abdul Khadar Syed for his help with SEM images.

DATA AVAILABILITY STATEMENT
Data from this research will be available upon request from corresponding author.