Pipeline Circumferential Cracking in Near-Neutral pH Environment Under the Influence of Residual Stress: Dormancy and Crack Initiation

Abstract

The purpose of this study is to identify the integrity challenges encountered by buried pipeline steels, specifically to address Circumferential Near-Neutral pH Corrosion Fatigue (C-NNpH-CF). Damage to the pipeline’s protective coating and corrosion conditions increase the risk of service failures caused by C-NNpH-CF. (Note that this mechanism has previously been termed near-neutral pH stress corrosion cracking.) Unlike axial cracking, circumferential cracking is primarily influenced by residual stress from pipeline bending, geohazards, and girth welds. External corrosion pits often lead to dormant cracks, with growth ceasing around 1 mm depth due to reduced dissolution rates. Investigating the impact of bending residual stress (an appropriate source of axial residual stress) and cyclic loading (simulated pipeline pressure fluctuation), the study employs the digital image correlation (DIC) method for stress distribution analysis. Factors like applied loading, initial notch depth, and bending conditions influence crack initiation and recovery from the dormancy stage by affecting stress distribution, stress cells, and stress concentration. Cross-sectional and fractographic images reveal time/stress-dependent mechanisms governing crack initiation, including dissolution rate and hydrogen-enhanced corrosion fatigue. The study emphasizes the role of various residual stress types and their interactions with axial cyclic loading in determining the threshold conditions for crack initiation.

1 Introduction

Circumferential Near-Neutral pH Corrosion Fatigue (C-NNpH-CF), which is a form of environmentally assisted cracking (EAC), has been shown to negatively affect the integrity and safety of pipelines transporting hazardous substances.^[1] Cathodic protection and protective coatings are employed for pipeline integrity maintenance to protect the pipe from corrosion caused by soil environments. Although these protective measures are in place, damage to coatings and corrosion cannot totally be avoided, so a safe design assumes some crack initiation in a narrow area within the disbonded coating where groundwater becomes trapped.^[2,3] This type of external pipeline failure involves the interaction of susceptible pipeline steel in contact with near-neutral pH ground water penetrating the disbonded coating under axial residual and cyclic stresses.^[4,5,6] The environment responsible for this form of transgranular cracking in the NN pH environment is a dilute carbonic acid solution (pH between 5.5 and 7.5) that forms when CO₂ from decaying organic matter in the soil dissolves in groundwater (in regions characterized by deciduous trees and monsoons, this impact occurs periodically), providing the necessary conditions for pipe external surface corrosion.^[1,7,8]

It should be noted that crack initiation and propagation mechanisms are thought to be different in NNpH environments.^[7] While EAC in near-neutral pH solutions is typically referred to as Near-Neutral pH Stress Corrosion Cracking (NNpHSCC), a number of studies^[4,9] have indicated that normal crack growth rates, especially in oil pipelines under monotonic tensile stress, are not high enough to cause concern even when the stress intensity factor (K) is considerable. In other words, significant crack growth has never been observed under static loading conditions except for very low rates in the initiation stage, whereas crack propagation requires cyclic loading.^[6,7] As a result, cyclic loading due to pipeline internal pressure fluctuations^[4] and diffusible hydrogen (a by-product of corrosion)^[9] are responsible for the corrosion fatigue mechanism of crack growth. The term Near-neutral pH Corrosion Fatigue (NNpH-CF) will be used hereafter to refer to what has been often previously referred to as NNpHSCC.

It is expected that cracks will be perpendicular to the maximum combined tensile stress, including internal pressures, axial stresses, residual stresses from pipe manufacturing and construction, as well as residual stresses from in-service damage.^[10] Axial cracking has become a serious and well-known threat to pipeline systems within the past decade.^[11] However, external cracking may occur at orientations other than axial, depending on the nature of stresses in the NNpH environment.^[11] In fact, when the total axial stress exceeds the total hoop stress, a gas or liquid pipeline can occasionally experience circumferential near-neutral pH corrosion fatigue, causing or contributing to failures.^[8,12,13] Many hazards and risks are assessed for and mitigated against axial cracks, but circumferential cracks are less well understood, and monitoring and mitigation methods are unclear.^[13,14]

The stress caused by internal pressure does not affect circumferential cracks as much as axially oriented cracks^[15] at least not in straight pipe sections. Nevertheless, the combination of axial cyclic stress from pipeline internal pressure fluctuations and other sources of axial stresses can reach a significant value to make the external surface of a pipe susceptible to circumferential crack initiation and propagation in NNpH environments.^[8] The following are some potential sources of axial stress:^{[4,8,12,13,16,17,18,19,20,21]}

• Hydrodynamic stress at bends

• Residual stress: thermal expansion, girth weld, pipe manufacturing, and pipeline construction like cold bend

• Axial stress from internal pressure

• Axial stress from geohazards: soil movement and settlement, freezing and thawing of soil, permafrost, road/ water crossings, impingement from rocks

• Bourdon effect on pipe bends

• Damage from manmade external sources like excavation

Corrosion fatigue of pipeline steels in the near-neutral pH environment is often modeled to occur in five stages: stage 0 (incubation stage involving coating deterioration and failure), stage 1a (crack initiation), stage 1b (early-stage crack growth), stage 2 (sustainable crack growth induced by mechanical driving force), and stage 3 (rapid crack propagation to rupture). Note that the sequence has been adopted to maintain the early historical numbering first introduced by Parkins.^[22] An important part of the pipeline’s lifecycle is stage 1b, which is crucial to pipeline integrity management.^[2,6,23] External corrosion can be caused by various mechanisms, including preferential dissolution at physical or metallurgical discontinuities like scratches, corrosion in stress raisers like pits, and galvanic corrosion related to microstructure or residual stresses. However, there is a decrease in dissolution rate as crack depth increases. This causes most initiated Stage 1a crack-like features to cease growing when they reach a depth of about 1 mm and enter the dormant stage.^[1,5,6,8] The majority of cracks in NNpH-CF colonies remain dormant, and only an estimated five percent continue growing.^[5,6,8] Indeed, reactivating a dormant crack requires certain location-specific conditions. Residual stresses near the crack tip and diffusible hydrogen could determine whether cracks are dormant or reactivated.^[8] In other words, it is possible for blunted cracks to be resharpened by the synergistic interaction of cyclic loading and hydrogen-related processes. Wherever these two conditions are met simultaneously, the probability of initiation from a dormant crack is higher.^[3,8]

While stress is a crucial parameter for this cracking mechanism, limited research has been dedicated to defining the role of stress type and value in dormancy and crack initiation at the notch (dormant crack) root. This is particularly relevant in cases with a stress gradient in the thickness direction, resulting from specific sources such as bending. This circumferential cracking can be considered a unique mechanism because the stress gradient in the thickness direction, along with other factors like initial notch depth (which influences stress concentration), dictates the cracking mechanism. This stands in contrast to a straight, smooth surface without any sources of in-thickness residual stress distribution.

This paper aims to investigate the stress distributions (derived from the digital image correlation (DIC) method) in the length and depth directions of pipelines under the simultaneous influence of bending residual stress (an appropriate source of residual axial stress) and cyclic loading (simulated pipeline pressure fluctuation). Following this, how stress distribution and depth of the initial dormant notch affect crack initiation will be addressed. The threshold condition for crack initiation will be determined by considering various types and amounts of residual stress and their interactions with axial cyclic loading. A deeper understanding of this mechanism may improve knowledge of Circumferential Near-Neutral pH Corrosion Fatigue (C-NNpH-CF), which plays an important role in pipeline integrity management.

2 Material and Experimental Methods

This section provides a comprehensive overview of sample preparation and parameter design procedures to rationalize correlations between laboratory experiments and real circumferential near-neutral pH corrosion fatigue (C-NNpH-CF).

2.1 Material Characterization

The material used in this study was a 16″ API-5L Gr. X52 with a wall thickness of 9.2 mm removed from a pipeline with a history of near-neutral pH CF cracking. The chemical composition of this pipe is given in Table I. Based on the literature review of this study, some information on the presence of C-NNpH-CF in pipelines and conditions for initiation and growth has been obtained by inspecting some in-service pipes removed from slopes.^[11,17] In most of these inspections, C-NNpH-CF happens close to the area where bending is an issue. As bending could be an appropriate source of residual axial stresses causing C-NNpH-CF in the field,^{[13,14,15,24]} in this study, bending is considered a significant source of axial residual stresses in the pipe. To investigate the interaction between bending residual stress and cyclic loading, these sections were not near or influenced by other residual stress sources like welding (seam and girth welds).

Table I Elemental Composition of the X52 Pipeline Steel

The actual tensile properties of the pipeline steels used for this study were determined in accordance with ASTM E8/E8M-22^[25] using 9.2-mm-thick rectangular sheet-type samples and a crosshead speed of 0.015 mm/mm/min. The test was performed using an INSTRON 8516 machine. Figure 1 shows the result of the tension test. This pipe’s yield stress (at 0.2 pct offset strain) and ultimate tensile stress were 356 and 561 MPa, respectively.

Fig. 1

Stress–Strain curve of X-52 steel used in testing

2.2 Specimen and Cell Preparation

The long samples are machined with the long tensile axis along the axial pipe direction of this pipe using electrical discharge machining (EDM). Figure 2 shows the design and dimensions of the samples.

Fig. 2

Design of long sample for C-NNpH-CF test (dimensions are in millimeters)

Three separate reduced gauge sections measuring 80 mm length × 20 mm width × 9.2 mm thickness with artificial through-thickness notches are explicitly designed to maximize the evaluated parameters. These three specified reduced-width areas in a long sample were bent at different degrees. It should be noted that the samples’ inner and outer surfaces must remain intact. Because cold working like machining can change the profile of residual stress distribution, the original curved surfaces were kept before bending.

Residual stress has a self-equilibrating feature, meaning that from Newton’s second law, the net force on an object at rest must be zero, so the sum of forces in any cross-section must be zero. As a result, if there is tensile residual stress on the outer surface of the pipeline in contact with the corrosive environment, the inner surface contains equivalent net compressive residual stresses and vice versa. Also, different bending angles could make various residual stress changes. Two bending angles of 10 and 20 degrees were designed to simulate cold bending in the field. For instance, Figure 3 illustrates a 20-degree bent pipe (each side is 10 degrees). This figure emphasizes that there are two different kinds of bending regarding the outer surface of the pipeline called inward and outward samples in this study. This feature simulates what happens to an actual bent pipe in the field since any bending that happens to a pipe because of slope or other reasons causes two recognizable bending directions regarding the outer surface of pipes. The distributions of residual stresses in these two conditions are more or less opposites, which will be discussed later.

Fig. 3

The 20-degree bent specimen. The central holes enable loading individual sub-samples for crack sharpening or testing if desired

Cold bending, as one of the causes of this type of cracking (because of pipeline construction or in-service damages related to geohazards, soil moving, etc.), creates axial stress in the whole bent area of the pipe which could be added to the cyclic loading from pressure fluctuation results in this type of cracking. It should be noted that there are two different conditions for a bent pipe. Figures 4 and 5 show the residual stress distribution in a bent section with and without spring back (y is the distance from the bending center on the pipe). As shown in Figure 4, if the pipe is not allowed to spring back, the axial strain would be D/2ρ on the convex side of the bend, where ρ is the bend radius and D is the pipe diameter. Allowing the pipe to spring back by removing the applied loads would alter the residual stress pattern, with the convex side going into compression and the concave side going into tension.^[17] While the first scenario can model a constrained dent or deformation condition of a pipe, the second can model an unconstrained dent, such as a rock or impingement being removed. As residual stress self-equilibrates, there is high tensile stress at the concave side to balance the high compressive stress at the convex side. It is possible to approach a material’s yield strength when adding this bending axial stress to other sources of stress.^[8,17]

Fig. 4

Stress profile from the concave to the convex side for the original bend. Reprinted with permission from Ref. [8]

Fig. 5

Stress distribution after the original bend is released. Reprinted with permission from Ref. [8]

Another source of axial stress is caused by pressure fluctuation in pipeline steel. Although internal pressure is the source of hoop stress in the pipeline, which expands the pipe circumference, this stress from internal pressure also pushes on the pipe ends and creates axial stress in the pipeline, as a pressurized pipeline transporting fluids usually has ends to press against. Typically, axial stress is about half as high as hoop stress.^[8,26] It is crucial to note that Bourdon and hydrodynamic stresses complicate what would happen in the field where a pipe bend is internally pressurized.^[8] The influence of these sources of axial stress on C-NNpH-CF crack initiation and growth should be explored in future research through more experiments and simulations.

According to the research done by Beavers et al.,^[10] pipelines operating at stress levels above 60 pct of the specified minimum yield strength (SMYS) are more susceptible to near-neutral pH corrosion fatigue. Therefore, 35 pct SMYS (approximately half of the hoop stress) and 50 pct SMYS (under the condition that some additional cyclic stresses could be added to the hoop stress) were considered for this study. Apart from that, this cyclic axial stress behaves differently in the inward and outward bending directions of the bent pipe, supposing no residual stress from the bend (or the bent section was heat treated and completely stress relieved). To clarify, the axial loading will make tensile stress on the concave side and compressive stress on the convex side of the bend. It would be similar to the residual stress distribution in a bent sample after spring back (without axial loading).

The initial crack/notch depth is another parameter to consider since the residual stress distribution close to the crack tip must be evaluated. An initial 1 mm depth notch was selected since many cracks stop growing after reaching a crack depth of 1 mm; at this point, they enter a state of dormancy.^[6] This observation is based on dormant cracks formed in straight pipes, so further investigation is needed for circumferential cracks formed in pipe bends. In addition, an initial notch with a 0.5 mm depth was added to the study to investigate the effects of initial notch depth on crack initiation. By increasing the initial notch depth to more than 1 mm, the stress intensity factor increases enormously, outside the scope of crack initiation. Therefore, the next step is to make artificial notches on each long sample’s bent parts. Figure 6 shows the position of the through-thickness notches on long samples (where X is the number of long samples, and 1–8 is the location of the notches on the long samples). The distance between notches is 20 mm (far enough from each other to inhibit the interaction of crack tip plastic zones close to the crack tip). A milling machine was employed to cut notches with a small width (about 0.2 mm) and various depths, as illustrated in Figure 6. This method provided rectangular-shaped notches to make stress concentrators at two corners. This would make controlling and evaluating crack initiation and growth at those corners easier.

Fig. 6

Schematic of artificial notch position on the long samples. (Width: 0.2 mm, through-thickness, distance between notches: 20 mm, notches are perpendicular to the surface of samples regardless of the bending angle and position of notches), The top is the outside surface

Furthermore, a specific cell (Figure 7) was designed as a container for C2 solution (synthetic field soil aqueous solution with a pH value of 6.3), which exhibits the greatest corrosion cracking potential, including 0.0274 gr. of MgSO₄.7H₂O, 0.034 gr. of CaCl₂.2H₂O, 0.0035 gr. of KCl, 0.0195 gr. of NaHCO₃, and 0.0606 gr. of CaCO₃ in each liter of deionized water) and the solution was sparged with a gas composed of 5 pct CO₂ and 95 pct N₂. The final pH of the solution was 6.3 ± 0.1, which is consistent with a near-neutral pH environment. Figure 7 shows how this particular cell is designed and how a long sample is placed therein. After transferring a long sample to the cell containing C2 solution and before starting cyclic loading, gas sparging with the rate of 1 bubble/second was performed for 12 days in a temperature-controlled laboratory (22 °C) to introduce the NNpH environment to the samples (the resulting corrosion leads to hydrogen diffusion into the sample).

Fig. 7

Specific cell designed for circumferential NNpH-CF

2.3 C-NNpH-CF Testing Conditions

This 12-day static corrosion test was followed by a C-NNpH-CF test (cyclic loading) with different axial loads, a frequency of 0.001 Hz and a stress ratio (Min stress/Max stress) of 0.1 on an Instron 8516 servo-hydraulic testing machine controlled with Wavemaker Runtime software. Since the stress was offset from the bending position, the applied stress was increased at bends by the bending moment. The test was conducted for 45 days unless a long sample failed during the test. Throughout the entire test, CO₂ was continuously sparged through the solution. According to the reasons that were provided in the previous sections, Table II shows the test plan to evaluate the effect of the mentioned parameters on the crack initiation mechanism.

Table II Test Parameters

Two-digit numbers were used to identify the notches. According to Figure 6, the first number indicates the long sample from which the notch originated, while the second number indicates the location of the notch on the long sample (1, 2, 6–8: inward bending direction, 3–5: outward bending direction). As a complementary test to long sample No. 1, long sample No. 2 was tested to evaluate the consistency of the data acquired. Two additional experiments were conducted on the non-bent sections having 0.5 mm and 1 mm initial notches, subjected to cyclic loading at 35 and 50 pct SMYS. The purpose of these tests was to evaluate the effect of cyclic loading on reference samples without any other sources of axial stress, such as axial bending stress. All other testing parameters (environment, cyclic loading pattern, etc.) were the same as in the previous test on long samples 1 to 8. For simplicity, these notches were designated as follows: A1 (0.5 mm, 35 pct), A2 (1 mm, 35 pct), B1 (0.5 mm, step 1: 35 pct for 45 days, step 2: 50 pct for 45 days), and B2 (1 mm, step 1: 35 pct for 45 days, step 2: 50 pct for 45 days).

2.4 Post-test Analysis

Upon completion of each test (45 days duration or fracture, whichever happened first), the long sample was rinsed with acetone and dried with hot air to prevent corrosion of the fracture surface, which might occur with water. Next, a precision saw fitted with a diamond-coated wafer blade capable of cutting widths of 0.2 mm was used to cut each unfailed notch into two samples (one half for cross-section analysis and the other half for fractography analysis) to characterize the depth profile of cracks and fractographic morphology.

The first half of each sample was ground to a 1200-grit finish and polished to a mirror finish using 6 and 1 μm diamond oil slurry and etched for 5 to 10 seconds using 5 pct nital (swabbed and then rinsed after 10 seconds) to characterize the crack location, depth, and morphology (cross-section characterization). The other half of each intact notch was intentionally fractured to investigate fracture surface morphology and crack growth behavior. The samples were placed in liquid nitrogen for up to four hours, removed, and fractured open by striking the backside notch with a cold chisel. As recommended in ASTM Standard G1-03, the fracture surface was dipped in a solution containing 6N HCl + 3.5 g/L hexamethylene tetramine to remove corrosion products. The fracture surface and cross-sections were imaged using a Zeiss Sigma 300-VP FESEM. A crack growth rate was calculated by dividing the measured crack growth depth by the number of cycles.

2.5 Stress Distribution Analysis via Digital Image Correlation (DIC) and the Calibration Method

Using digital image correlation (DIC), strains are measured through a non-contact procedure from images of a sample under tension in 2D or 3D. This technique can be applied to various materials research and component testing applications to analyze specimen static and dynamic behavior. In many cases, strain gauges are replaced by DIC strain maps.^{[27,28,29,30]} Appendix A provides further details on digital image correlation (DIC) and the calibration method.

3 Results

3.1 Stress Distribution Analysis Based on the DIC Method

To determine the stress distribution in the external surface and thickness direction of the bent gauge area from all sources of stresses, two long samples bent at 10 and 20 degrees were loaded to 35 pct SMYS and 50 pct SMYS to develop the strain map in gauge length using the digital image correlation (DIC) method. This method was employed on long samples without notches to simplify the effect of axial stresses on circumferential cracking. However, it is obvious that the presence of notches gives more precise data, which could be the subject of further research.

Figure 8(a) illustrates the final strain distribution in three bent gauges (the center one is in the outward bending direction, and the left and right ones are in the inward bending direction) and two grip regions, which make a distance between the bent gauges. Different regions on this graph show the level of strain: inward bent centerlines have the maximum tensile strain (maximum level of positive strain), though the center of outward bending direction at the middle-bent gauge has the maximum negative/compressive strain. Also, it could be seen that strain decreases with distance from the bend centerline. Finally, grip regions show the strain/ stress distribution based on the cross-section area under loading.

Fig. 8

(a) strain distribution on the long sample external surface under 50 pct SMYS cyclic loading (20-degree bending), (b) strain distribution in the thickness direction (inward bending direction: external surface in tension, internal surface in compression)

Figure 8(b) shows an in-thickness directional strain map for the bent grip with an inward bend. Type I residual stresses, which vary continuously over at least several grain diameters, must be self-equilibrated such that they sum to zero in any direction over the total cross-section. Because they sum to zero over the cross-section, it is expected that the highest compressive residual stress gradient in the depth direction accompanies the highest tensile residual stresses.^[3,31] This feature is shown in Figure 8(b), where the highest tensile strain/stress distribution in an inward 20-degree bending gauge length on the external surface is accompanied by the highest negative strain/stress on the internal surface. Since the thickness of bent samples is the same, maximum stress on the sample results in maximum stress reduction in the thickness direction, too.

3.1.1 Level and nature of final stresses in the inward bending direction

Figure 9(a) presents the normal stress distribution along the bend centerline (between − 20 and +20 mm of the bend centerline) at the surface and the notch root (0.5 mm far from the external surface) of long sample No.6 (20-degree bending, 50 pct SMYS, 0.5 mm initial notch depth). The stress level at the position of the notch root is less than the external surface because of the stress self-equilibrium feature. It should be mentioned that as a result of the stress magnification of the notch, the actual stress at the notch root is a totally different matter. An additional study examined the effects of maximum cyclic loading on stress distribution and crack initiation. Figure 9(b) shows the final stress distributions for different notches (combined residual stress from 10-degree bending and SMYS loading of 50 and 35 pct: Samples No. 3 and 5).

Fig. 9

(a) Stress level at the surface and bottom of pits in various initial notch positions (sample No.6: 20-degree bending, 50 pct SMYS, 0.5 mm initial notch depth), (b) Final stress distribution in 10-degree bending, 50 and 35 pct SMYS cyclic loading: (samples No. 3 and 5)

The graph in Figure 10 depicts stress distribution in the width direction (the width component of the bending stress vector) on the external surface of a 10-degree inward bending direction (line A-B in Figure 8(a)). According to standard ISO 7539-5, there is a stress gradient through the width of the gauge sample, varying from edges to the center, i.e., in the inward bending direction, the tensile stress is lower at the edges than at the center.^[32] It is consistent with the stress distribution shown in Figure 10 (10-degree bending, 50 pct SMYS, 0.5 mm initial notch depth), where the maximum stress level at the edges was 235 MPa compared to the 245 MPa at the center.

Fig. 10

Normal stress distribution in the width direction: 10-degree bending, 50 pct SMYS, 0.5 mm initial notch depth (line A-B in Fig. 8(a))

3.1.2 Level and nature of the final stresses in the outward bending direction/non-bent sections

The outward bending direction is affected by two sources of compressive residual stress, one from bending after spring back and the other from axial cyclic loading, as shown in Figure 8(a)’s central section of the long sample. These stress distributions can be affected by other testing parameters, such as the bending angle. As an example, Figure 11 illustrates the stress distribution in the outward bending direction of long samples No. 6 and 7. In both graphs, compression stress was shown at different values depending on how far the initial notch was from the bend centerline. The bending residual stress generated by 20-degree bending and 10-degree bending is the reason for the huge difference between samples No. 6 and 7 stress distribution. Axial cyclic loading is 50 pct of SMYS for both mentioned samples.

Fig. 11

Compressive stress level in various initial notch positions under axial cyclic loading of 50 pct SMYS: (sample No.6: outward direction/20-degree bending), and (sample No.7: outward direction/10-degree bending)

Finally, the stress distributions were evaluated for the non-bent sections. These sections did not have any bending, which means the only source of stress was axial cyclic loading, which was a maximum of 50 pct SMYS (180 MPa), 35 pct SMYS (126 MPa) at the maximum tensile load, 5 pct SMYS and 3.5 pct SMYS at the minimum level, according to the ratio of minimum to maximum load of 0.1 in every corrosion fatigue test cycle.

3.1.3 Crack initiation and early-stage crack growth at initial notches under tensile stresses

To examine crack initiation from notches, selected notches were sectioned. For instance, Figure 12 shows cross-section and fractography images of crack initiation from the bottom of the artificial notch (No. 61: 20-degree bending, 0.5 mm initial depth, 50 pct SMYS cyclic loading, 10 mm from bend centerline) at different magnifications. This study assumed that pitting happened at the first stage of the SCC/CF mechanism in the NNpH environment, and these artificial notches simulate the pits or dormant cracks.

Fig. 12

Crack initiation from the bottom of notch No. 61 (20-degree bending, 0.5 mm initial depth, 50 pct SMYS cyclic loading): (a) cross-section of the initial notch, (b) fractographic images

This figure shows that at the beginning, the microstructurally short cracks can initiate at the bottom of a notch and then grow to the size of 0.1 to 1 mm which is related to mechanically short/ shallow cracks (Early-stage Crack). In general, it should be mentioned that crack initiation in near-neutral pH environment is not a fatigue process, and previous attempts to categorize it as such were not successful.^[4,9] It involves, in fact, various galvanic processes such as stress cells caused by coupling the regions with different nature of residual stresses (compression vs. tension), dissimilar metal cells formed between cracked mill scales and underneath steel substrate, and between pearlite bands and ferrites, and even concentration cells due to the concentration gradient of CO₂ and resulting species of corrosion of CO₂.^[4,5,9] The specific location of crack initiation at the notch root is influenced by metallurgical factors such as the presence of non-metallic inclusions, grain boundaries, and specific phases (pearlite/ferrite or mill scale). Microstructurally short cracks, formed due to the galvanic nature of corrosion, are typically found at the interface between the ferrite phase and the pearlite colony, as outlined in References 1, 5, 33, 34. According to Figure 12(a), this crack initiation occurs in the first several grains and the rest of the region is early-stage crack growth which is transgranular. The next section will discuss factors that can affect crack initiation and growth to mechanically short/ shallow cracks.

As illustrated in Figure 12(a), in the first stage, crack propagation occurred in the area with the highest shear stress, responsible for plastic deformation.^[23] Since the stress intensity factor on the main crack plane is the highest, it is unlikely that the crack will propagate in a direction not perpendicular to the maximum principal stress. Branched cracks may be forced back to the main crack plane, leading to a “zig-zag” pattern on fractured surfaces.^[35]

The crack initiation was confirmed by fracture of the sample at the notch root site. According to the fractography morphology of notch No. 61 (Figure 12(b)), the fracture surface after placement of the sample in liquid nitrogen is divided into three regions: the initial notch region results from milling (upper part), crack initiation (middle part), and the final intentionally fractured regions. The middle part exhibits quasi-cleavage with areas of smooth brittle fracture and ductile dimples.^[36] It should be noted that all fracture surfaces were subjected to ongoing corrosion since the sample was immersed in an NNpH environment during the test. Thus, the surface quality is affected by this corrosion. The fine features that would show up in quasi-cleavage in a dry hydrogen atmosphere will therefore be subject to corrosion particularly if they are exposed to the solution for an extended length of time. In the bottom part, a complete cleavage region is typical of brittle fracture in liquid nitrogen, similar to cleavage fractures in low-carbon steels under impact below the transition temperature.^[36]

Table III provides an overview of average crack growth derived from fractography images over the entire width, specifically in the inward bending direction with tensile residual stress. Consequently, the following section explores how these parameters influence crack initiation at the bottom of an initial notch. It is crucial to note that this study exclusively concentrates on crack initiation and its influencing factors. The analysis of long cracks, where fracture mechanics come into play, will be thoroughly addressed in another part of this investigation. This study excluded failed notches as well.

Table III The Crack Growth in Various Notches Under the Influence of Different Testing Parameters (Inward Bending Direction)

3.1.4 Effect of tensile stress on crack initiation and early-stage crack growth

Referring to Figure 9(a) (normal stress distribution along the bend centerline (within − 20 mm and + 20 mm of the bend centerline)), Figure 13 illustrates the crack depth observed at notches No. 61, 62, 66, and 68, which were located at − 10, + 10, − 20, and + 20 mm from the center in the gauge section with inward bending, respectively. The sample was tested under 50 pct SMYS for 354 hours/1274 cycles. Notches 61 and 62 (10 mm far from the bend centerline) have a crack depth of around 50 μm (at the edge), although notches 67 and 68 (20 mm from the bend centerline) show less than 10 μm crack depth. This correlation confirms the effect of stress on crack initiation, as more stress results in more crack growth.

Fig. 13

(a) crack depth in notch No. 61 (− 10 mm), (b) crack depth in notch No. 62 (+ 10 mm), (c) crack depth in notch No. 66 (− 20 mm), (d) crack depth in notch No. 68 (+ 20 mm)

It is important to keep the bend angle and residual stress from bending constant throughout the experimental procedure in order to evaluate the effect of stress from cyclic loading on crack initiation. According to Figure 9(b), the same 10-degree bend in sample number 3 with 50 pct SMYS produced a higher stress level than a 35 pct SMYS loading in sample number 5. Figures 14(a) and (b) illustrate crack depths caused by these various cyclic loadings (at the same location: − 10 mm far from the bend centerline). As a result, the crack depths at the bottom of notch numbers 31 and 51 are 50 and 15 μm, respectively. It is consistent with the stress level at the bottom of the pits (notch number 31: 150 MPa, notch number 51: 70 MPa).

Fig. 14

(a) crack initiation in 50 pct SMYS cyclic loading, (b) crack initiation in 35 pct SMYS cyclic loading (other parameters kept constant: 10-degree bending, 1 mm initial depth, 10 mm from the bend centerline)

Figure 15 illustrates cross-sectional images of notches No. 31 and 71. Both notches are 10 mm from the 10-degree inward bend centerline with 50 pct SMYS, but initial notch depths for notch numbers 31 and 71 are 1 and 0.5 mm, respectively. The duration of both tests is almost the same. Notch No. 71 cross-section shows a crack growth of less than 10 μm, but the crack depth on notch number 31 cross-section is more than 50 μm, so it seems that the effect of crack initiation governed by the dissolution process is less significant than the effect of mechanical processes. The stress concentration factor provides more comprehensive data on crack initiation in the depth direction. Despite a decrease in stress level in the depth direction of a bend section, a higher depth increases the stress concentration factor. As a result, the maximum stress level at the bottom of the 1 mm initial notch is higher than the maximum stress level at the bottom of the 0.5 mm initial notch.

Fig. 15

Effect of initial notch depth on the crack initiation (a) notch No. 71: 0.5 mm initial notch depth, (b) notch No. 31: 1 mm initial notch depth (other parameters kept constant: 10-degree bending, 50 pct SMYS, 10 mm from the bend centerline)

3.1.5 Crack initiation and early-stage crack growth along specimen width

Notch 77, listed in Table III and illustrated in Figure 16, could be used to examine the crack initiation behavior along the specimen width. The gauge section containing Notch 77 was bent by 10 degrees, and the stress distribution in the width direction is shown in Figure 10, from which the difference in stress between the edge surface and the mid-width is about 10 MPa. The cross-sectional images 16a and b show the bottom of notch No. 77, which contains the initiation of the crack (sampling location is defined in Figure 8(a)). The crack initiated from the notch root at the edge surface of the specimen is short but remains relatively sharp, while the crack from the notch root located at the mid-width is long but has a wide crack crevice and blunt crack tip. The latter morphology is consistent with dormant cracks often seen in near-neutral pH SCC cracks.^[8]

Fig. 16

Crack initiation at the edge and center of the width of notch No. 77 (10-degree bending, 50 pct SMYS, 0.5 mm initial notch depth), (a) crack initiation at the edge: section A on Fig. 8(a) (b) crack initiation at the center of width: section C on Fig. 8(a)

The different crack behavior and morphologies observed at the edge surface, and the mid-width could be related to the magnitude of stresses in the material surrounding the notch root. Also, the edge surface is in plane stress where the material at the notch root is under biaxial loading and is subject to stress comparable to the yield strength, while the material at the notch root in mid-width is in plane strain, and the stress in every direction could be as high as 3 times the yield strength because of the presence of triaxial stresses. This high triaxial stress can of course attract hydrogen. Consequently, the risk of initiating a crack from the notch root at mid-width is much higher. However, the triaxial stress zone surrounding the notch root exists over a limited distance to the notch root, leading to limited crack growth, probably within the triaxial zone. Further growth is prohibited because of the sharp reduction of stresses beyond the triaxial stress zone. Under the circumstances, corrosion is continuous on the crack walls and at the crack tip, which blunts the crack tip and increases crack crevice because of the non-passivating nature of near-neutral pH solutions. In contrast, crack initiation from the notch root at the edge surface is limited to a few grain sizes, a typical dimension of a microstructurally short crack. In near-neutral pH solutions, these microstructurally short cracks are formed along grain boundaries and at interfaces between pearlite colony and ferrite phases, which appear consistent with the observations shown in Figure 16(b). Because of the narrow crack crevice found in Figure 16(b), the initiation of microstructurally short cracks could occur at a much later testing time. According to Figure 16(b), initiation occurs at both corners at first, but after a while, one corner grows faster than the other. Indeed, stress shielding subsequently prevents the other one from growing, which means crack growth consumes the stress in one corner and there is then not enough stress for the other one to grow.

3.1.6 Crack initiation and early-stage crack growth in different test durations

Figure 17 illustrates crack depth versus the test duration in different notches with various parameters (notch No. 16: bending angle of 20 degrees, 50 pct SMYS loading, notch No. 36: bending angle of 10 degrees, 50 pct SMYS loading, notch No. 46: bending angle of 20 degrees, 35 pct SMYS loading) at the same location (inwards, 20 mm from the bend centerline). The test durations varied because these long samples (1, 2, 4) failed at different notches. As a matter of fact, these notches could experience a different number of cycles/times in the test. Section IV will discuss how to consider all relevant factors simultaneously (initial notch depth, axial loading, and bending angle). Nevertheless, only three notches are shown in Figure 17 to explain the concept behind the effect of test duration on crack initiation.

Fig. 17

Crack initiation as a function of test duration: (a) crack depth vs. time of test, (b) crack depth at notch No. 16 (20 degrees, 50 pct SMYS), (c) crack depth at notch No. 46 (20 degrees, 35 pct SMYS), (d) crack depth at notch No. 36 (10 degrees, 50 pct SMYS)

As can be seen in this figure, after 750 cycles/208 hours (test duration for long sample No. 1), notch No. 16 does not crack despite having the highest level of stress (95.8 MPa from bending residual stress and axial applied stress). Notch number 46, with a lower stress level (70.4 MPa. because of 35 pct SMYS loading instead of 50 pct SMYS in notch number 16), indicates a crack depth of about 30 μm. The maximum crack growth compared with other notches (50 μm) is observed in notch 36 (Figure 17(d)) after 3051 cycles/848 hours with stress of 66.5 MPa, although the average crack depth in the whole width is about 0.2 mm. The figure shows that crack initiation requires a definite time at the bottom of the pit, even though the stress level is high. Accordingly, this supports the hypothesis that different factors affect this time. Despite the presence of severe conditions, such as maximum tensile stress, which results in more hydrogen diffusion to the plastic zone close to the crack tip and more stress concentration at the notch corners (more corrosion in the presence of stress cell), a minimum time is required for hydrogen to diffuse to this plastic zone, and for corrosion to occur at stress concentrators at the bottom of the pit.^[5]

In Figure 18, partial crack initiation is shown as a feature of crack initiation within a limited time (less than the minimum time needed for crack initiation). A full fractography image of notch number 46 (20-degree bending, 35 pct SMYS, 1 mm initial notch depth) shows regions without crack growth (about 30 percent of the width). It is worth noting that there is a time and location-dependent process for developing microstructural cracks at the notch root.^[3,8] The weakest link sites, such as inclusions near the notch root, can facilitate crack initiation. As discussed, stress distributions in the width direction can also influence partial crack initiation. Under the condition that this notch had been kept in the test longer, i.e., had threshold time passed completely, more uniform crack depth would have been observed (as shown in other crack initiation figures). Therefore, constant crack growth requires surpassing some minimum time, which is specific for each condition. In fact, crack initiation in the whole width should be subject to a lower and upper threshold. When the lower threshold is exceeded, microstructural cracks appear somewhere in the width direction, while the upper threshold results in cracks appearing throughout the sample’s width.

Fig. 18

Partial crack initiation (Notch No. 46: 20-degree bending, 35 pct SMYS, 1 mm initial notch depth)

3.2 Crack Initiation and Early-Stage Crack Growth at Notches in the Outward Bending Direction (Compressive Stresses) and Non-bent Sections (No Residual Stresses)

A summary of average crack growth based on fractography images, specifically in the outward bending direction with compressive residual stress, is found in Table IV. Almost all gauge lengths experiencing compressive stress on the surface (and also at the notch roots), including notches (X3, X4, and X5, where X is the number of long samples 1 to 8), do not show any crack initiation. Indeed, regardless of the bending angle, location of notches with respect to the bend centerline, level of cyclic stress, and initial notch depth, there is almost no crack initiation in the outward bending direction.

Table IV The Crack Growth in Various Notches Under the Influence of Different Testing Parameters (Outward Bending Direction)

The general agreement is that compressive residual stress suppresses crack growth, while tensile residual stress promotes it.^[23] Previously discussed in Figure 11 as a sample of compressive residual stress distribution, the outward bending direction is simultaneously affected by two distinct sources of compressive stress: residual compressive stress from bending (after spring back) and compressive stress from cyclic loading.

There was only one exception for the mentioned explanation: Notch No. 64. Among all notches experiencing compressive stress, this notch had the highest level at the bottom of its notch. This is related to the 20-degree bending angle, bend centerline position, and 50 pct SMYS cyclic loading (Figure 11). It is evident from Figure 19(a) that a tiny crack initiation with a depth of 5 μm happened at both corners of this notch at the outer edge of the sample (shown in section D in Figure 8(a)), where the stress concentration is high. The tiny cracks are located at the interface between the ferrite phase and the pearlite colony, reflecting the formation of microstructurally short cracks caused by the galvanic nature of the corrosion, as detailed in Reference 33. It is also consistent with the study by Van Boven et al., emphasizing that surfaces with compressive residual stress can show little pitting.^[5] However, there is no chance for this tiny crack-like feature to propagate since a positive high tensile residual stress gradient must be present for mechanically driven crack growth to occur beyond a critical distance.^[5,31,37] Figure 19(b) shows the notch root morphology from the sample’s mid-width (shown in section F in Figure 8(a)), indicating that crack initiation under the test conditions is limited and must be microstructurally related if it occurs.

Fig. 19

Crack initiation under compressive stress at notch No. 64: 20-degree bending, 50 pct SMYS, 0.5 mm initial notch depth, outward direction (a) crack initiation at the outer edge of the sample: section D on Fig. 8(a) (b) crack initiation at the mid-width of the sample: section F on Fig. 8(a)

After passing the steps of cycling loading on the non-bent section on notches with different initial depth/cyclic loading (A1, A2, B1, B2), only a short crack around 60 μm was observed on the cross-section of notch B2 (1 mm initial depth, 35 and then 50 pct SMYS cycling loading), according to Figure 20(a) and Table V. Partial cracking was evident on the fractured surface (Figure 20(b)). In this case, partial crack growth results from axial cyclic loading without other sources of axial stress like bending. Other non-bent section notches provided less stress at the notch root (stress × stress concentration, to be discussed later), so Table V summarizes the results for the non-bent section.

Fig. 20

Partial crack initiation (Notch B2: non-bent section (1 mm initial notch depth/35 and then 50 pct SMYS) (a) cross-section analysis (SEM- BSD) (b) fracture surface

Table V The Crack Growth in Various Notches Under the Influence of Different Testing Parameters (Non-bent Section)

As mentioned before, the non-bent sections have no residual stress gradient in the thickness direction. The stress concentration factor changes as the initial notch depth increases from 0.5 mm to 1 mm. Moreover, the dissolution mechanism cannot cause crack initiation for these notches. As a result, circumferential crack initiation generally cannot occur in a pipeline with only axial loading (from pipe internal pressure fluctuation) and no other sources of axial loading, like bending residual stresses.

4 Discussion

Generally, three stages are involved in initiating short cracks: formation of pits, initiation of microstructurally short cracks, and growth of microstructurally short cracks into mechanically short cracks.^[31,34]

The same process of crack development at the bottom of a pit could be considered at the stage after dormancy and crack initiation from a dormant crack:

• 1. Crack initiation and microstructural short cracks depended on galvanic processes, microstructure, and stress concentrators.

• 2. Growth of mechanically short/ shallow cracks with the dimension from 0.1 to 1 mm, above which fracture mechanic principles can be applied, although crack growth rate for these cracks is more than for long cracks under identical mechanical loading conditions in terms of stress intensity factor.^[3,38]

These are microstructurally short cracks of a size equal to the grain size or other microstructure within the material, usually less than 100 μm. Fracture mechanics does not normally adequately describe the growth of these cracks.^[31,34] Note that initiating microstructurally short cracks is also possible under constant stress loading.^[3]

Dissolution could result in crack initiation if the dissolution rate in the depth direction is at least ten times higher than in the lateral direction.^[5] Where crack initiated at the notch root depends on metallurgical factors like the presence of non-metallic inclusions, grain boundaries, specific phases (pearlite/ferrite or mill scale)^[1] and mechanical factors like stress raisers and stress cells.^[34]

In another mechanism of crack initiation at the bottom of the pit, mechanical process, microcracks are initiated at the weakest link sites in the hydrostatic zone ahead of a notch tip, not exactly at the corner of the notch. Compared to high pH SCC,^{[39,40,41,42,43,44,45,46]} this is a completely different case. Accordingly, a micro-crack emerges due to a sharp crack tip propagating back to the bottom of the pit, resulting in a sharp crack forming there. There are two types of weakest link sites: metallurgical discontinuities with low bond strength or regions with high hydrogen segregation.^[5]

This investigation deals with the initiation of cracks from notch roots of pipeline steels exposed to a typical near-neutral pH environment that can cause pipeline failure through the mechanism of hydrogen-enhanced fatigue. The notch has a depth of either 0.5 mm or 1.0 mm and was made in the pipe wall with a varied distribution of residual stresses, either compressive, tensile, or free of residual stresses introduced by bending. The width of the notch was controlled within 0.2 mm as the width of cracks found in pipeline steels from field operation can reach 0.5 mm in width.^[47]

The initiation of cracks in pipeline steels exposed to near-neutral pH environments has been primarily investigated on straight steel surfaces that are either polished, mill-scale covered, or have a varied distribution of residual stresses.^{[23,33,34,48,49,50,51]} It has been determined that crack initiation on a straight external surface often starts with pit-formation, followed by crack initiation from the bottom of the pits.^[5] During this stage, coating conditions, soil conditions, and steel metallurgy all play a role. Mechanical driving forces, such as operating pressure fluctuation, are less significant.^[6] An important point to remember is that initiation is a corrosion process driven by stress cells and heterogeneity of the structure, which allows chemical potentials to vary.

One of the common characteristics of cracking in pipeline steels exposed to a near-neutral pH environment is the phenomenon of crack dormancy, a situation in which cracks often cease to grow when they reach a depth of about 1.0 mm. While evaluating dormancy, it is critical first to understand why dormancy occurs, which includes but may not be limited to the conditions summarized below:^[6,8,31]

• 1. A primary cause of crack dormancy is related to the reduction in dissolution rate at the crack tip. This is due to a complicated process that involves CO₂ gradients and ionic concentration variation in the aqueous solution in the pit.

• 2. The nature and magnitude of residual stresses at and near pipeline steel’s outer surface significantly affect dormancy, specifically when residual stresses dominate a stress distribution in the thickness direction. For instance, decreasing local stresses in the thickness direction as the crack grows may hamper the onset of stage 2. Most crack colonies do not propagate beyond 1 mm because mechanical forces are below the threshold for crack propagation.

• 3. Hydrogen acts together with residual stresses, and if the hydrogen concentration in the material surrounding the crack tip is low, the crack will be more likely to remain dormant.

• 4. Crack blunting caused by room-temperature creep may also retard crack growth.^[8] The additional deformation from this creep can sometimes increase residual compressive stresses ahead of a crack tip and enhance the retardation effect. In other words, cyclic creep retardation is related to the cyclic hardening effect. Cracks can grow when sustainable strain exists at their tips, so applying pressure slightly higher than operational stress could exhaust this strain and retard crack growth.

A superposition of dissolution growth curves and hydrogen-facilitated fatigue growth curves is shown in Figure 21 for crack initiation and growth of near-neutral pH SCC/CF of pipelines.^[23] During crack initiation and early-stage crack growth, the former is predominant, while the latter results from fatigue and hydrogen combined dominates in the latter stages. A major role in crack growth in stage 1b should be played by dissolution; hydrogen-facilitated fatigue growth is less evident during this phase because of the benign mechanical driving force. To extend service life, it is crucial to understand how a dormant depth tip becomes active for stage 2 growth.^[6,23]

Fig. 21

Schematic illustration of the effect of $\Delta K^{\alpha } K_{{{\text{max}}}}^{\beta } /f^{\gamma }$ upon SCC velocity of pipeline steel exposed to near-neutral pH aqueous bicarbonate solution. Reprinted with permission from Ref. [22]

The initiation of cracks from a dormant crack has been a subject of research over the last decades. To reactivate a dormant crack, certain conditions must be met. The first one is the magnitude and nature of the stresses near the crack tip. The second relates to the amount of diffusible hydrogen reaching the crack tip. There is a connection between these two factors.^[2,3,6,8] As mentioned before, sustainable crack growth is possible if two extrinsic sources of crack sharpening, i.e., tensile cyclic loading and the hydrogen enhancement of fatigue, can prevail over crack tip blunting and dormancy.

For crack reactivation, minimum stress must exist ahead of the crack tip. Growth, dormancy, and reactivation of blunted cracks may occur repeatedly in a growing crack. The combination of hydrostatic stresses in the plastic zone, hydrogen segregation, and the presence of weakest link sites, such as inclusions near the crack tip, can cause hydrogen-induced microcracks to form, resharpening the main crack. It is only possible for cracks to grow by this mechanism if they are subjected to cyclic loading. Indeed, hydrogen-related sharpening and cyclic loading can synergistically resharpen blunted cracks. There is a high probability of growth wherever these two conditions coexist.^[3,8,34] These factors dominate the crack initiation of blunt notches in this investigation. Crack tip morphology, either sharp or blunt, is another important factor affecting the crack growth rate, which is not considered in this research, so all initial notches were blunted.

With the above introduction, the initiation of cracks from notch roots in the current investigation is much like crack initiation from dormant cracks. Various situations, including the magnitude of externally applied stresses and the residual stresses introduced by bending, and varied depths of notches have been created in the present investigation to understand the mechanisms of crack initiation from the notch roots.

4.1 Crack Initiation and Early-Stage Crack Growth from Notch Roots in the Absence of Tensile Residual Stresses

For the notch roots embedded within the pipe wall that were not processed to introduce tensile residual stresses, the initiation of cracks from notch roots appears very difficult compared to the notch roots with tensile residual stresses. This seems surprising, as the actual stresses at the notch roots can be very high, although the applied nominal stresses were low, 35 to 50 pct SMYS. Based on the stress concentration factors determined by Neuber’s Rule,^[52] the actual stresses at the notch roots are determined to be 415.8 and 458.6 MPa for notches with an initial depth of 0.5 and 1.0 mm, respectively, when the applied nominal stress is 50 pct SMYS.

At both notch depths, the actual stresses at notch roots exceed the yield strength of the pipeline steel. Under this magnitude of stresses, cracks can be readily initiated on smooth surfaces, either freshly polished or covered by mill scales. It has been determined that the rate of crack growth in the stage of crack initiation is well above 10⁻⁸ mm/s on a straight and smooth surface,^[34] which is about an order of magnitude higher than the crack growth rate calculated based on the rate of general corrosion of pipeline steel coupons exposed to the same near-neutral pH solution. This indicates there must exist a galvanic process during crack initiation. On a straight and smooth surface, a galvanic cell is formed between the mill scale and the steel under a cracked mill scale, causing localized corrosion such as pits. Under the circumstances, cracks could be initiated only when such pits are located within the material with high tensile residual stresses and can be extended to a deeper depth if the tensile residual stresses can be maintained in the thickness direction. On a polished surface, galvanic corrosion is found along the interfaces between pearlite bands and ferrite.

The crack growth rate at the root of notch B2 in Table V is determined to be 7.7 × 10⁻⁹ mm/s, which is lower than the crack growth rate during crack initiation on a straight and smooth surface but still much higher than the crack growth rate calculated based on the rate of general corrosion of pipeline steel coupons exposed to the same near-neutral pH solution. It suggests that galvanic corrosion still plays a role in crack initiation at the root of notch B2.

The galvanic cell responsible for crack initiation at notch B2 is a stress cell coupled between the locations with different tensile stresses, such as the corner of the notch root and the wall of the notch. The difference in potential between the anodic surface and the cathodic surface would govern the corrosion rate. Pitting corrosion and pit-induced cracking were not found at the notch root with a depth of 0.5 mm, indicating insufficient potential difference because of the lower stress difference between the anode and cathode. In the case of notch B2 (1 mm depth), pitting corrosion and pit-induced cracking were found in the width section away from the edge surface because of slightly higher stress. As a result, the loading condition applied to notch B2 would represent the threshold condition of crack initiation for the notch root (notch B2) with a notch depth of 1.0 mm.

4.2 Crack Initiation and Early-Stage Crack Growth from Notch Roots/Dormant Cracks in the Presence of Residual Stresses

The presence of residual stresses changes the magnitude of the difference in potential between the anode and the cathode and their location with respect to notch geometries.

When notches are made in the specimen with compressive residual stresses, the notch roots are under compressive stress even when a tensile stress of 50 pct SMYS is applied. This would make the surface of the notch wall that is free of loading an anodic site and the notch root a cathodic site. This is a typical situation of large anode (notch wall) and small cathode (notch root), which would prevent any corrosion and crack initiation from the notch roots.

When notches are introduced in the specimen with tensile residual stresses, the difference in stress and, therefore, the potential between the anode/notch root and the cathode/notch wall is increased, depending on the magnitude of tensile residual stresses. This is a typical situation of a large cathode (notch wall) and small anode (corner of notch root). Under the circumstances, pit-formation and pit-to-crack initiation at the notch root become possible.

However, by examining cracks found at the notch roots with tensile residual stresses, as shown in Figure 22, the typical scenario of pit-formation and pit-to-crack initiation, as seen in the case of crack initiation at the notch root free of residual stresses, is not observed. This may reflect how tensile residual stresses affect the process of crack initiation.

Fig. 22

(a) a blunt tip resembling a notch root, (b) a blunt tip crack with an indication of corrosion, (c) a sharp crack that is usually advanced by the mechanism of hydrogen-enhanced fatigue. Reprinted with permission from Refs. [8, 53]

In the case of crack growth in near-neutral pH solutions, three types of crack tip morphologies, as shown in Figure 22, are often observed.

• (a) A blunt tip resembling a notch root, as shown in Figure 22(a), is a common morphology found in dormant cracks with reasons, as explained previously,

• (b) A blunt tip crack with an indication of corrosion, as those observed in the current investigation. Under the circumstances, crack walls are separated because of corrosion and crack tips have a radius well larger than 10–100 atoms, which is the radius of a crack defined by fracture mechanics,

• (c) A sharp crack is usually advanced by the mechanism of hydrogen-enhanced fatigue, in which corrosion plays little role in crack advancement. This often occurs when crack growth is relatively high and/or crack growth is under cathodically protected conditions.

The absence of pitting corrosion at the notch roots with tensile residual stresses is caused by insufficient time of corrosion, very localized anodic sites, or both. For those notch roots with a crack growth rate at the level of 10⁻⁸ mm/s, the time of corrosion available at the notch roots is an order of magnitude less than that available at the notch root without tensile residual stresses. Upon the initiation of a crack from the notch root, the anodic site shifts to the crack tip and further corrosion of notch roots is prohibited. The anodic site would also be restricted to the direction with the highest shear stress. As soon as the crack is formed, further corrosion of the crack wall would be limited because the anodic site is always associated with the crack tip.

4.3 Threshold of Crack Initiation and Early-Stage Crack Growth under the Influence of Residual Stresses at the Notch Root

Figure 23 summarizes the hypothesis mentioned above regarding the impact of accumulated stress (bending residual stress and applied loading) on the average crack growth rate occurring at the notch root (dormant crack). As previously stated, a minimum stress requirement exists at the notch root for crack initiation, which was determined to be about 85 pct SMYS. Below this stress threshold, a crack growth rate at a level of 2 × 10⁻⁹ mm/s was also observed, as shown in Figure 23. This crack growth rate is equivalent to the crack growth rate calculated from the general corrosion rate of flat coupons exposed to the same NNpH solution.^[54] Under the circumstances, accelerated crack initiation and growth arising from stress cells would be unlikely if it existed at all.

Fig. 23

Crack growth rate as a function of maximum stress at the notch root

To confirm the benign impact of a stress cell, the two data points with a growth rate comparable to the crack growth rate calculated based on the rate of dissolution measured using flat coupons were further examined. They were from notches No. 56 and 58, with a bend angle 10 deg and notch depth of 1.0 mm under an applied stress of 35 pct SMYS. The two notches were also located 2 cm from the bend centerline and would have the lowest tensile residual stresses and applied stresses. Only some microstructurally short cracks with dimensions of about 1-grain size were observed, as shown in Figure 24.

Fig. 24

SEM images showing microstructurally short cracks initiated from the notch root of notch No. 58

5 Conclusion

This study analyzed circumferential crack initiation at the bottom of dormant cracks/pits in steels exposed to near-neutral pH environments. The following are some key findings of this investigation.

• 1. The digital image correlation (DIC) method could be considered an appropriate technique for C-NNpH-CF investigations since this method can analyze final stress distribution affected by various sources of residual and applied stresses. In this study, final stress distribution emphasizes that the accumulation of stress from axial loading and residual stress increases the level of stress (compressive/tension) at the bending centerline since the type of axial stress and residual stress in the inward bending direction is tension and in the outward bending direction is compression, regardless of the bending degree and the level of axial loading.

• 2. The initiation of microstructurally short cracks from the bottom of the pit may result from localized galvanic corrosion at ferrite/pearlite interface at the locations with the highest tensile stress levels. Corrosion in the form of stress cells at the notch corners (stress concentrators) is responsible for this crack initiation mechanism.

• 3. The presence of a high level of stress at a stress cell would suffice for crack initiation at the notch root by direct dissolution, regardless of the direction of stress (compressive/tensile), emphasizing that a shallow notch with high compressive residual stress can show little crack initiation at stress concentrators. However, these tiny cracks cannot propagate.

• 4. Stress gradient in the depth direction is a significant feature of circumferential NNpH-CF. Therefore, the stress × stress concentration factor provides more comprehensive data on crack initiation in the depth direction since it considers both effects of stress gradient in the depth direction and notch depth simultaneously.

• 5. A tensile stress gradient (difference in potential between the anodic surface and the cathodic surface) specifies the first step in the initiation of cracks at the bottom of the pit/dormant cracks. At the crack tip, localized anodic sites inhibit the formation of further corrosion of the notch root, and the crack wall results in blunt tip cracks. It is possible to have a sharp crack when the hydrogen-enhanced fatigue mechanism dominates.

Appendix A

An overview of the method employed in this study to find the stress distribution in the width and thickness directions of bent samples using the DIC method is presented in this section, along with a calibration method defined to ensure the results of stress acquired by the DIC method are accurate.

The same long sample has been used for C-NNpH-CF tests (with 10- and 20-degree bending angles) employed for the digital image correlation technique. A thin white layer was painted to prepare the long samples for the DIC system, and a black speckle pattern was applied with spray paint before the DIC system was used. Afterward, the specimens were loaded up to a maximum force of 22.8 kN (35 pct SMYS of X52) and 32.6 kN (50 pct SMYS of X52), respectively. The test was conducted on a computer-controlled MTS tensile testing machine with a maximum load capacity of 30 MN. Force and crosshead displacement were recorded simultaneously during the test. In addition, sequential images of the gauge region (inward and outward bending directions) in length and thickness directions were captured at each 100 N increase in force using a high-resolution digital camera for DIC measurements.

A commercial DIC software, Vic-3D 9, was used to analyze sequential images obtained during the test. First, strain maps on external pipe surfaces and thickness surfaces were derived from DIC methods. After that, the stress–strain curve (introduced in Section II–A) was employed to determine stress distributions along the length, width, and thickness directions of gauge areas. The procedure to convert strain to stress includes defining a desired point, reading strain values from DIC export data, and finally finding the equivalent stress related to that strain from the stress–strain curve.

A DIC algorithm calculates displacement and strain by comparing a series of gray-scale images of a sample at various stages of deformation, tracking pixel movement in the Region of Interest (ROI), and using a correlation algorithm. First, a discrete function describes the gray level of each pixel in the image, and then a correlation is calculated for each pattern. This principle states that the displacement of the spots accurately represents specimen deformation and strain.

For example, Figure A1 illustrates a long sample’s external surface strain map containing three 20-degree bent gauge lengths (two sides are inward, and the center one is outward). What is required for this study is finding stress levels at different points of interest related to the corrosion fatigue mechanisms. Therefore, the calibration method should be defined here to ensure the result of stress acquired from the mentioned procedure. To achieve this goal, the Region of Interest (ROI) should be found far from any source of plastic deformation, like edges, holes, and bent areas. Therefore, the best ROI could be in the grip region far from the central hole and edges. Based on the DIC report, the strain of the selected ROI in Figure A1 is 0.000393. Since the grip region is just under 50 pct SMYS load, elastic modules of steel (about 210 GPa) gave us a stress value of around 82 MPa.

Fig. A1

Digital image correlation (DIC) calibration method

On the other hand, based on the grip region size (45 mm × 9.2 mm) and load applied to the sample (32.6 kN equivalent to 50 pct SMYS of API 5L X52 pipeline steel), the stress value was calculated around 84 MPa. As a result, this method seems to be considered a calibrated way to find stress distribution in a bent sample under the influence of two main axial sources of stress: residual stress from bending and axial loading tensile stress. This minor difference could be explained by the fact that the DIC experimental technique used for this long sample analysis provides a method with this level of accuracy. Plastic deformation sources at the selected ROI may also be responsible for these small errors.