Full-Scale Testing of Corrosion Resistant Alloy-Mechanically Lined Pipes for Submarine Pipelines

The requirement to specify corrosion-resistant alloy (CRA) material for submarine pipelines arises from the need to resist the internal corrosion caused by the transported fluids under conditions in which it is not technically, operationally, or economically feasible to use carbon steel. One primary corrosive agent is carbonic acid which is formed by the dissolution of carbon dioxide into water [1–7]. The rate of corrosion generally increases with the increase in the temperature and partial pressure of carbon dioxide. The selection of an appropriate CRA is made more complex by the presence of hydrogen sulfide [8,9].

The motivation for the current work stems from the need to ensure the long-term reliability of corrosion-resistant alloy-mechanically lined pipes (CRA-MLPs) in submarine applications. In such applications, the triple point is a known area of vulnerability, where the combination of material interfaces and cyclic loading can lead to fatigue failures. Despite its critical importance, there is a lack of comprehensive data and guidelines specifically addressing the fatigue performance of the triple point in the existing standards such as API-5LD [10] and DNV-ST-F101 [11]. Consequently, this study aims to fill such a gap by conducting full-scale resonance fatigue testing (FSFT) on CRA-MLP, with a focus on the triple point. By simulating the real-world operational conditions that such submarine pipelines face, the current study seeks to provide valuable insights into the fatigue resistance of CRA-MLPs. The scientific interest in this work lies in its potential to inform future updates to industry standards, thereby enhancing the safety and reliability of subsea pipeline operations. The results of this study also contribute new data that not only address a critical area of concern but also advance the understanding of the fatigue behavior of CRA-MLPs.

The fatigue performance of CRA-MLPs, particularly at the triple point, has been the subject of limited studies in the literature. Previous research has primarily focused on the fatigue behavior of CRA-clad pipes, with few studies addressing the unique challenges posed by mechanically lined pipes. Standards such as DNV-RP-C203 [12] and BS 7608 [13] provide general guidelines for the fatigue design of CRA-MLPs but do not explicitly address the triple point. Additionally, there is limited availability of the S–N curves specific to the triple point, making it difficult to accurately predict fatigue life in such a critical region. This study thus aims to fill this gap by providing detailed fatigue performance data for the triple point, obtained through full-scale bending and resonance fatigue testing.

In the subsea pipelines, the triple point, transition, or seal weld is defined as the weld that lies between the CRA liner, host/backing pipe, and weld overlay cladding, and this weld can be classified as a dissimilar weld (see Fig. 1). Currently, the widely recognized approach for fatigue qualification is the new welding procedure in which fatigue performance on specimens containing welds is verified by full-scale resonance testing. The objective of this testing is to confirm that the fatigue performance of the transition or seal weld is in line with the existing standards design S–N curve.

The objective of the CRA liner in the MLPs is to provide the required protection for the backing steel (i.e., carbon steel) against the corrosive environment. Given this matter, the CRA liner must remain intact for the entire design life of the pipeline. Any damage in the CRA liner may lead to direct contact between the corrosive fluids and the backing steel and consequently corrosion and pitting [14,15]. A qualification program for CRA-MLPs was undertaken by Tkaczyk et al. [16,17] on a 10.75-in. outer diameter (OD) pipe with a total wall thickness of 25.4 mm, lined with Alloy 625 (UNS N06625 [18]). This pipe was subjected first to reeling simulations followed by full-scale fatigue testing. The pipe did not show any sign of cracks at the weld overlay/liner transition point and the fatigue performance was above the DNV class D target using the OD stress range.

Another full-scale fatigue testing [19] was conducted on 219.1 mm OD pipes lined with 3 mm alloy 625. The full-scale fatigue testing was undertaken after the pipes were subjected to reeling simulation with internal pressure. The tests were stopped at DNV class F1 target life and the internal surface was inspected with no sign of damage detected. Additional full-scale testing [20] was carried out on 219 mm OD pipes of carbon steel thickness of 13.5 mm with 3 mm of CRA liner. In the tests, three specimens were subjected to coating simulations, followed by pressurized reeling simulations and then full-scale resonance fatigue testing. In these tests, all three specimens exceeded the DNV class C target using OD stress ranges. Some specimens were tested to failure and cracked at the DNV class B1 target life. The crack was first initiated on the outer surface of the carbon steel before cracking occurred at the triple point. DNV JIP [21] conducted full-scale fatigue testing on two mechanically lined pipes at a nominal stress range of approximately 150 MPa. The outer diameter of the pipes was 323.8 mm with a backing steel thickness of 17.5 mm and a liner thickness of 3 mm of 316L. The fatigue tests were terminated due to fatigue cracks that were initiated from the cap weld toe on the outside. Fatigue lives just above 1 million cycles were recorded in both tests which were slightly above the mean D-curve in the DNV-RP-C203.

Putra et al. [22] also explored the fatigue behavior of the CRA-MLPs, particularly focusing on the seal welds. Their study employed small-scale low-cycle fatigue tests to evaluate the fatigue life of seal welds under high-pressure, high-temperature (HPHT) conditions. While this approach provided valuable insights into the behavior of specific components, it did not encompass full-scale fatigue testing of the entire CRA-MLP system. The focus on small-scale fatigue tests, such as those conducted by Putra et al. [22] leaves a significant gap in the literature regarding the comprehensive fatigue performance of CRA-MLPs, particularly at the triple point.

Small-scale tests often fail to account for modifications in stress distribution and constraint near the triple point, which is influenced by the interaction of the backing pipe, liner, and seal weld. These features can cause stress concentrations that are challenging to detect at the scale of small specimens, reducing the likelihood of accurately representing pre-existing flaws. Furthermore, the effects of bending, thermal gradients, and axial forces during real-world pipeline operations amplify these localized stress variations, which are critical for understanding fatigue and failure mechanisms. Consequently, the lack of full-scale testing addressing these nuances creates a gap in ensuring the reliability of CRA-MLPs in HPHT environments. This highlights the critical importance of full-scale fatigue testing in replicating the complex interactions of stresses and constraints within the pipeline system, providing a more accurate evaluation of the triple point's fatigue behavior and overall system reliability.

There have been concerns over the effect of interface flaws at the liner-to-weld overlay transition, particularly during the reeling installation [23]. Therefore, an engineering critical assessment (ECA) of interface flaws may be required, as indicated by Kaspary et al. [4]. Based on the ECA outcomes [17], it was evident that the typical interface flaws were revealed not to be critical for the integrity of the CRA-mechanically lined pipes during the reeling process and the subsequent service during the design life. When assessing the criticality of any defect, the height (through-thickness dimension) is the most important dimension. As DNV-RP-C203 [12] neither specifies the triple point location nor recommends the most suitable fatigue class for the triple point, full-scale testing is often executed to conservatively determine the pipe fitness for a specific application. Ultrasonic inspection (UT) of the triple point should be avoided largely because of concerns about its effectiveness. According to Tkaczyk et al. [17], flaws may be present at the triple point from the fabrication process, and potentially grow during installation and subsea operation. These “features” occur mostly due to the combination of the residual stress in conjunction with the formation of the deleterious phases close to the dissimilar metals, particularly in the triple point.

The focus of the present paper is to present the results of a full-scale resonance fatigue testing that was undertaken on a 10-in. CRA-MLP with a 3 mm liner. The liner in the testing was made of CRA 625 (nickel-chromium) alloy. The pipeline was installed using S-lay to tie back the subsea manifold with a floating facility. The information obtained from the testing can be used in the pipeline standards and codes to properly classify the most suitable fatigue class for the triple point. Another motivation for this paper is that there is limited published data on the fatigue performance of CRA-mechanically lined pipe 625-weld overlay and this paper provides such information which can be used by practitioners and engineers who are performing structural integrity assessment for the 625-weld overlay. However, it should be emphasized that the results from the four specimens presented in the paper are not sufficient to help practicing engineers as the number of tested specimens is not enough to be directly transferable. It should also be noted that the gripping force, which is the ability of the CRA liner to resist wrinkling due to the higher radial contact stress between the liner pipe and the outer pipe, is negatively influenced by the thermal cycle associated with some pipe coating processes. This phenomenon occurs because the thermal cycle can reduce the residual compressive stress that maintains the mechanical bond between the liner and the outer pipe, potentially leading to a decrease in the gripping force. However, during the pre-commissioning phase, the presence of the system pressure associated with the hydrostatic test is likely to result in a substantial increase in gripping force again. Consequently, to reproduce the condition of the pipeline that will be put into service, the CRA-mechanically lined pipe was subjected to a simulated coating cycle at 250 °C ± 10 °C. The simulated coating cycle was carried out by an induction heating coil used in a coating yard as shown in Fig. 2.

The methodology for the current study was designed to comprehensively evaluate the fatigue performance of CRA-MLPs, with a particular focus on the triple point. The testing program was carried out in two sequential phases, as follows:

• Full-Scale Bending Test (FSBT): Specimens were subjected to installation-like stresses using a four-point bending rig at CLADTEK Batam. This phase simulated the strains experienced during S-lay pipeline installation, effectively pre-straining the specimens to reflect real-world operational conditions.

• Full-Scale Fatigue Test (FSFT): Following FSBT, the pre-strained specimens were transported to TWI Cambridge, where they were exposed to cyclic loading to replicate the fatigue conditions expected during pipeline operation. This phase was critical to assess the fatigue resistance of the triple point under operational stress cycles.

By adopting this above two-step testing strategy, the current study ensures that the fatigue performance of the CRA-MLPs is evaluated in conditions closely resembling the actual lifecycle of submarine pipelines. The integration of FSBT and FSFT phases provides a robust foundation for understanding the interactions between installation-induced strains and operational fatigue stresses at critical locations such as the triple point.

There are three common types of CRA subsea pipes: (i) solid CRA pipes; (ii) CRA-clad (metallurgically bonded) pipes; and (iii) CRA-mechanically lined pipes. The least costly option out of these three types of subsea pipes is very often the CRA-mechanically lined pipe. It has been used in the oil and gas industry for more than 40 years for transporting corrosive fluids and presents an excellent track record. For some projects, and considering their overall life cost, CRA-mechanically lined pipes are more cost-effective compared to CRA-clad pipes, solid CRA pipes, and even in some cases carbon steel pipes with corrosion inhibitors. Other advantages of CRA-mechanically lined pipe in offshore pipelines as compared to CRA-clad pipes are listed in the following [24–27]:

• Improved fit-up due to tailored made ends, resulting in superior installability and reliability—a project offshore Australia reported 2.5 times faster installation in the CRA-mechanically lined pipe section versus the CRA-clad section, with the same installation procedure and vessel.

• A wider choice of CRA compared to the clad plates—as heat treatment restricts the workable envelope. Carbon steel and CRA materials are heat treated using their optimal parameters.

• A wider range of diameter/wall thickness ratios compared to CRA-clad JCO/UOE forming processes.

• Generate almost 50% less CO₂ compared to similar CRA-clad pipes—CRA-mechanically lined pipes are also easily recyclable.

• Generate reduced lead delivery times compared to CRA-clad or solid CRA pipes.

On the other hand, the disadvantages of the CRA-mechanically lined pipes can be summarized as follows:

• As the liner is not metallurgically bonded to the carbon steel outer pipe, there is an additional failure mechanism of liner buckling—normally referred to as “wrinkling”—during installation or strain associated with lateral buckling. Despite this, CRA-mechanically lined pipes have been successfully installed in the subsea by reeling on several projects [16–24] and many are currently in service without concerns.

• Shorter range of diameters as compared to CRA-clad pipes which typically range from “8 to 60,” whereas mechanically lined piped ranges from “6 to 36.”

The three techniques for manufacturing CRA pipes can be summarized as follows:

•    Solid CRA pipes: These are manufactured by processes similar to those of carbon steel viz. seamless and longitudinally welded (SAW). The solid CRA materials used in offshore pipelines are duplex stainless steel and weldable martensitic stainless steel.

•    Metallurgically bonded pipes: These are produced using a heat and deformation process to bind the internal clad layer to the outer shell of carbon steel. The most common route is forming the pipe from the clad plate. There are three techniques to produce the clad plates namely: (1) hot-rolling bonding; (2) explosive bonding; and (3) weld overlaying. Hot roll-bonding accounts for more than 90% of clad plate production worldwide. Hence, the hot-rolling bonding will be discussed in this paper rather than the other two techniques. The hot-rolling bonding method is outlined in the following steps:

  • Preparing a “sandwich” of carbon steel-CRA/CRA-carbon steel plates;

  • Seal welding the edges of the “sandwich”;

  • Hot rolling to reduce thickness and increase surface area;

  • Removing seal welds;

  • Surface finishing the CRA layer.

The clad plate is then formed into a cylinder by brake press and the longitudinal seam is welded in the typically using submerged arc welding (SAW).

•    Mechanically lined pipes: These are manufactured by the following steps:

  • Forming a thin wall corrosion-resistant liner pipe and inserting the corrosion-resistant liner within a carbon steel outer pipe (see Figs. 3 and 4). In this case, the CRA pipes are not bonded to the outer pipe but rather held by friction grip. For the CRA-lined pipes, the most common CRA liners that have been installed include 316L, 317L, 904L, 6Mo, Incoloy 825, Inconel 625, and 22 Cr duplex stainless steel. It is imperative that the yield strength of the backing steel exceeds that of the CRA liner or else the backing steel will deform plastically before the liner and the liner will retreat from the backing steel upon relaxation of the hydraulic pressure. The most used backing steel grades used in the oil and gas industry are DNV 450 (X65) and DNV 415 (X60). There is no practical restriction for using higher grade backing steels, being the limiting factor often the overmatching of the pipeline girth welding when CRA fillers are required.

  • Subjecting the composite pipe to a hydraulic internal expansion process, which is a common method to achieve bonding between the carbon steel pipe and the CRA liner. In this process, the carbon steel pipe remains in the elastic regime, while the CRA liner deforms plastically. The elastic recovery of the carbon steel pipe generates residual compressive stress in the liner, creating a mechanical bond. Alternative expansion techniques, such as mechanical expansion or thermal expansion, can also be employed. For example, heating the outer pipe creates differential expansion, increasing the contact pressure, and enhancing the bond between the layers.

  • Inserting the CRA liner tube into the outer carbon steel pipe, after which the liner and outer pipe are placed into a die to control the dimensions of the outer pipe. The liner is then expanded using hydro expansion until it contracts against the outer carbon steel pipe. At this point, both the liner and outer pipe are further expanded, resulting in the plastic deformation of both layers. However, this technique has certain downsides. Non-standard outer carbon steel pipes must be procured to ensure that the final pipes meet standard size requirements. Additionally, the process introduces the Bauschinger effect, which can reduce the material's yield strength during subsequent loading cycles, potentially affecting the structural integrity and long-term performance of the pipe.

  • Applying a cladding weld overlay to extend the CRA layer to the pipe ends, and replacing the earlier reliance on seal welding the CRA liner to the carbon steel pipe. This approach is now standard practice and provides a continuous, corrosion-resistant layer across the entire pipe length, ensuring better alignment during installation and reducing risks associated with weld defects or gaps. The use of cladding weld overlay enhances both the installability and the long-term performance of the pipeline by improving joint integrity and minimizing potential weak points.

For any pipeline system, fatigue damage may accrue from the loads incurred by the following items: installation, spanning (e.g., vortex-induced vibration and direct wave action [26]), thermal cyclic loading, system effects (e.g., slugging [28,29]), and on-bottom pipeline instability [3]. However, the most significant fatigue damages are associated with thermal cycling, spanning by vortex-induced vibration and installation, hence, will be briefly explained in the following.

• Thermal cycling: During the operation stage of the pipeline, it is expected that the pipeline will experience cyclic changes in temperature and pressure due to planned and unplanned shutdowns and subsequent startups. The changes in stress that occur at the location of lateral buckles during these shutdown and startup events are likely to have fatigue damage implications.

• Spanning by vortex-induced vibration: Pipelines installed on an uneven seabed tend to form free spans rather than follow the topographical features of the seabed. When these spans get exposed, an external flow will shed vortices in their wake. These vortices cause local pressure variations on the surface of the cylinder. Under certain conditions, such pressure variations will be sufficient to cause the pipeline to oscillate. The frequency at which these vortices are shed depends on the velocity of the flow and pipe diameter. If the shedding frequency approaches the natural frequency of the pipeline, a condition called “lock-in” occurs. This is where the shedding frequency rapidly collapses to the natural frequency of the pipe span. This resonant condition causes large amplitude oscillations, which can rapidly accumulate fatigue damage on the pipeline.

• Installation: The accumulated installation fatigue damage is usually determined by evaluating strain cycles induced by vessel motions. The suspended pipe length and average lay rate are taken into account and the vessel motions are based on the worst-case sea state and wave heading for pipe lay operations. For pipelines installed by S-lay, a stress analysis should be performed to determine the installation stress histogram and the installation fatigue demand the pipeline would experience during installation—in these cases, normally the maximum stress occurs when the pipe is on the stinger. The installation stress histogram can be developed using either of the following options: (i) assume a constant stress range at each girth weld location equal to the maximum stress range at that location; or (ii) perform a rain flow analysis on the stress versus time response. During pipe clamping (e.g., for in-line structure installation, pipeline abandonment, or recovery during reel-lay or J-lay), significant fatigue damage can be accumulated.

The triple point is defined as the intersection of CRA liner, weld overlay, and backing pipe (refer to point C in Fig. 5). The region CD (Fig. 5) is a complex joint as it lies between three materials (host pipe, liner, and weld overlay) and is generally understood as the weak joint for the CRA-mechanically lined pipe. The main functions of the clad overlay weld, as shown in Fig. 5, are to hold the liner in position, seal the annulus, and prepare for girth welding including by allowing for fine tailoring of the CRA-mechanically lined pipe end-tolerances. Any lack of fusion or disbanding between the clad weld overlay and the backing steel could threaten the integrity of the CRA-mechanically lined pipe. The girth weld between the pipe ends/joints (typically performed on the installation barges or spool-bases) is typically made of CRA and should have sufficient quality (i.e., free of flaws such as lack of fusion, cracking, and porosity). Due to the coarse-grained structure of the weld metal and geometric constraints associated with the weld overlay, the inspection of these welds becomes more difficult, requiring special techniques and validations. Figure 6 presents examples of cracking at the triple point introduced following a full fatigue resonance testing carried out by TWI [30].

There are gaps in the current standards and codes regarding the manufacturing of the CRA-mechanically lined pipes and the qualification protocols required to verify their integrity. For instance, API-5LD [10] and DNV-ST-F101 [11] do not define the triple point. However, in the DNV-ST-F101, Section D.8.11.5, there are acceptance criteria for the seal weld. In the authors' view, the DNV-ST-F101 is still silent regarding the following important items:

• No established criteria for addressing the potential flaws at the triple point, such as lack of fusion and laps, which are known to significantly influence fatigue performance. Additionally, there are no workmanship-type criteria that explicitly link the geometrical and microstructural features of the triple point to its fatigue behavior. This absence of standardized guidelines or inspection methodologies leaves a critical gap, making it challenging for designers and manufacturers to reliably ensure the integrity and performance of this crucial region during both fabrication and operation.

• No definitions or comments regarding the length of weld overlaid ends. 100 mm is the typical length required in the industry as it allows for beveling preparation and girth weld cut whilst providing sufficient distance between the girth weld toe and triple point to avoid overlapping stress concentrations. Cost savings associated with shorter weld overlays are worth exploring during the design phase. It would therefore be worthwhile defining conditions for allowing shorter weld overlay lengths—particularly after the cutouts. Furthermore, although CRA-mechanically lined pipes are similar to CRA-clad pipes in many ways, additional concerns related to the integrity of the triple point are applicable, and this is understood to be its most sensitive region. Hence, it is crucial to implement stringent control during the manufacturing of the CRA-mechanically lined pipes to ensure a high-quality triple point is obtained.

It should be noted that neither DNV-ST-F101 [11] nor DNV-RP-F108 [31] requires an ECA for the triple point. A major consequence of this matter is that, although automated ultrasonic testing (AUT) must be applied to the non-destructive testing of the girth welds, radiographic testing (RT) is the applied technique for triple points. It only permits sizing defects by their length and width, while AUT allows for defect sizing in three dimensions. When assessing the criticality of any defect, the height (through-thickness dimension) is the most important dimension. Ultrasonic inspection of the triple point is still avoided largely because of concerns about its effectiveness. The experience of AUT on projects involving CRA-clad items concerning some projects has been disappointing [32]. Some highlighted issues include the increased rate of welding repairs associated with false calls, the time required to determine whether there was an indication with the weld, and the challenges posed by CRA materials being noisy and anisotropic. Combining RT and phased array ultrasonic testing (PAUT) has proven to be an effective approach to disambiguate false calls, particularly when RT alone may result in subjective assessments due to the bi-dimensional nature of the images. PAUT offers additional insights by providing depth and volume data, allowing for better sizing and characterization of potential flaws. This complementary use of RT and UT techniques represents a practical solution to improve inspection reliability for CRA welds, ensuring enhanced accuracy in identifying true defects. As such, the CRA materials degrade the ultrasonic wave propagation resulting in difficulties in the interpretation of the signals. DNV-RP-F108 [31] is also silent regarding any recommendations for conducting project-specific small-scale testing to determine the tensile properties, fracture toughness, and fatigue crack growth. Furthermore, DNV-RP-F108 [31] does not provide any guidance on how to prepare a segment specimen for the triple point including the dimensions of the specimen.

Another challenge associated with the triple point is the characterization of different microscopic features and flaws initiated during the manufacturing process, as they may grow further during the installation and in-service phase as a result of the operational loads [32]. As the CRA layer is thin (normally 3 mm), the current available non-destructive inspection has limited success in detecting such microfeatures. It needs to be noted that the interface between the liner and the substrate pipe prevents the transfer of sound from the base metal to the liner.

It is thus recommended that the industry prioritize the development of advanced inspection techniques capable of detecting features or flaws smaller than 0.5 mm in the triple point region. Ongoing efforts, such as the TWI Joint Industry Project (JIP) “Development of Product Qualification Workmanship Acceptance Criteria for Mechanically Lined Pipe—Phase 2,” are addressing this need. These initiatives focus on characterizing triple point flaws and establishing workmanship criteria through FSFT, providing a pathway to standardizing qualification processes and enhancing the reliability of mechanically lined pipes in demanding applications.

In the current study, the testing program was executed across two specialized facilities to ensure comprehensive pipeline qualification. The FSBT at Cladtek Batam introduced installation-like stresses through S-lay simulation, preparing the specimens for subsequent evaluation. These pre-strained specimens were then tested at the TWI facility in Cambridge under controlled cyclic loading to assess the fatigue performance of the triple point and other critical areas. Details of the testing conducted are described in detail in the following sections.

This section describes the methodology, objectives, and results of the FSBT. This test simulates the stresses experienced during the S-lay installation of submarine pipelines, particularly focusing on the triple point's response to these stresses. Detailed explanations of the test setup, loading conditions, and strain measurement techniques are provided herein. The primary goal is to introduce the stresses and strains that the pipe would encounter during installation, ensuring that the test conditions closely replicate real-life scenarios. As part of the full-scale fatigue testing program in the current study, a full-scale bending test (i.e., installation simulation) was conducted on the test strings fabricated from a nominal outside diameter of 10.75 in. (273.05 mm) made of DNV seamless 450 backing steel with a wall thickness of 25.4 mm, lined with 3 mm of CRA nickel-chromium alloy 625. The typical objectives of the full-scale being tested or the S-lay simulations are to

• Demonstrate the capacity of the CRA-lined pipes to withstand the bending strains likely to occur during the installation at ambient temperature and pressure.

• Illustrate that the CRA-lined pipes can withstand reverse cyclic bending during installation.

It should be noted that the main objective of the installation simulations undertaken before the full-scale bending testing was to introduce the stress and the corresponding strain representing those that the pipe would experience at the pipelay vessel, before the subsequent full-scale fatigue testing (i.e., pre-straining the specimens). The global strain was used in this work and calculated from the curvature of the pipe between the two load points, measured using a “wire potentiometer” displacement transducer. The deflection measurement was monitored during the test and was used to determine the global strain to which the pipe was exposed. It should also be indicated that the test string had two girth welds and four triple points, and the length of the test string was 11.2 m. Figure 7 shows the four-point testing rig used to simulate the S-lay installation in the neutral position (Fig. 7(a)) and in the loading position at +0.4 global strain (Fig. 7(b)). Figure 7(c) illustrates the pipe in the loading position at −0.4% global strain (reverse loading). Each pipe specimens were loaded as per the loading history shown in Fig. 8. Three cycles were applied (three cycles with +0.4% global strain and three reverse cycles with +0.4% global strain). The number of cycles mimicked the laying processes and any associated abandoned and recovery. From the video cameras used to inspect the inner surface after completing the loading cycles required for the S-lay simulation, there was no sign of wrinkles or ripples (see Fig. 9). It should be noted that although the objective of the bending test was to introduce stresses and strains to replicate the likely expected installation loads and not to check the wrinkling of the liner. The pipes were inspected to confirm that there were no wrinkles formed during the installation simulations.

This section covers the FSFT that follows the FSBT. This test aims to evaluate the long-term fatigue performance of the triple point under cyclic loading conditions that replicate those encountered during pipeline operation. The methodology includes the application of cyclic bending moments to the specimens, to determine the fatigue life at the triple point. The section below discusses the target S–N curves, stress ranges, and the specific conditions under which the tests were conducted. Testing was conducted under controlled, room-temperature conditions with cyclic loading frequencies representative of submarine pipeline operations. While environmental factors such as sour service were not simulated, the cyclic bending applied allows isolation of the triple point's fatigue response to mechanical loading alone. By establishing a controlled testing environment, the study provides a reproducible baseline for evaluating the fatigue resistance of the triple point, which serves as a foundational step for future assessments under more complex operational scenarios.

The four test strings, which were bent using the four-point bending test rig, were cut from the ends. This was to achieve the required dimensions, as shown in Figs. 10 and 11, to suit the requirements of the full-scale fatigue test rig. It is obvious from Fig. 12 that the pipe had a residual curvature which evidences that the pipe was previously bent to simulate the installation process. When the pipe was loaded in the full-scale fatigue test rig located at the TWI facility, the liner seam weld was located at the 3 o’clock position due to the gravity effect and the “banana shape” of the pipe. Figure 11 shows the strain gauges that were instrumented on each specimen. Each specimen was instrumented with two rings of eight single-element strain gauges equally spaced around the circumference of the pipe, on each side of the girth welds. So, each test specimen had 32 strain gauges in total. The strain gauges were positioned 80 mm from the center of the girth weld. In other words, the strain gauges were located 60 mm from the weld toe and such distances were selected to ensure that the strain gauges were aligned with the triple points.

The method of leak detection used to determine the liner cracking before the through-wall crack developed during the full-scale fatigue was drilling two holes. The two holes were 5 mm in diameter and were drilled through the carbon steel thickness only. Ultrasonic testing was undertaken to measure the actual thickness of the carbon steel at the designated hole location—the location of the two holes can be seen in Fig. 11. Additionally, an absorbent paper was utilized by placing the paper under each drilled hole. Then this paper was connected to the test machine via wires to trip the circuit and stop the test in case of any liner cracking before the through-wall crack was developed. The test specimens were with water and pressurized to 594 bar. A special colorant was mixed with the water so that in the case of any leakage from the drilled holes due to the failure of the liner could be differentiated from moisture from other sources. This pressure was applied to produce an axial mean stress of 1000 bar based on the nominal pipe dimensions. Also, the internal pressure was applied to ensure that the mean stress in the cycle was always positive. In other words, the R ratio was greater than 0. Figure 13 shows an example of the strain range measured by the different strain gauges. The same figure shows that the R ratio applied during the full-scale fatigue tests was above 0 and in the range of 0.6.

All specimens were tested in air at ambient temperature using the resonance fatigue machine. This machine imposed a rotating alternating bending moment at a frequency of 30 Hz. The test speed was controlled to achieve the required stress range for the test. The machine was adjusted to ensure the average of the strain gauge reading corresponded to the target stress range. The number of test specimens is presented in Table 1. The inner wall stress range was calculated by multiplying the measured stress range on the OD by 0.792 and this factor is the ratio of the inner diameter (ID) to the outside diameter.

The full-scale fatigue test aimed to determine whether the liner triple points/weld overlay cladding can qualify for DNV class (E) on ID and DNV class (D) on OD with a survival probability of 97.7% and a minimum confidence level of 95%. The appropriate DNV-RP-C203 [12] S–N curves for the component girth welds are as follows:

• Weld root—either E, F1, or F3 curve is appropriate dependent on the level of misalignment (hi/lo) achieved during pipelay.

• Weld cap—the D-curve is the most appropriate curve for all weld caps.

Based on the above equation, the factor of 1.645 was obtained from the probability tables, assuming a normal distribution, for a probability of 95%. Hence, the target life was a factor of 3.283 times the DNV class (D) and class (E) design curves. It should be noted that for the number of cycles on the ID (class E), the inner wall stress range was used. It should be that the triple point is a non-standard geometry. TWI was running JIP between 2012 and 2014 to establish the fatigue performance of CRA-mechanically lined pipes [33]. The JIP concluded from the endurance test conducted that the design curve of the results was mid-way between DNV class (D) and DNV class (E) (using OD stress ranges). For these reasons, DNV class (E) on ID and DNV class (D) on OD were selected.

For uncracked triple points, the stress range was determined based on the average readings from eight strain gauges positioned around each triple point. In contrast, for uncracked girth welds, the nominal stress range was calculated using the average readings from 16 strain gauges—8 positioned on each side of the weld, as illustrated earlier in Fig. 11. Due to the uneven stress distribution caused by residual ovalization from full-scale bending tests, the test parameters were conservatively calibrated to reflect the lowest stress measurement recorded among all strain gauges. This approach ensures a conservative and reliable assessment of stress ranges for both triple points and girth welds.

Intermediate dye penetrant inspection (DPI) was performed on the whole length of the clad weld overlay as well as each weld overlay/liner triple point after reaching the class (E) target on the ID. In this event, the pipes were drained of water and were saw-cut at one end of the test specimen to remove the end cap. This was done to easily access the inside of the pipe. However, before the dye penetrant inspection, the interfaces between the liner and the clad weld overlay were cleaned using soft tissue paper. Once this was done, the intermediate dye penetrant inspection was completed using qualified technicians. Concerning the post-test inspection, upon the completion of the tests, the specimens were drained of the water. Then for each test specimen, the sections of the pipe containing the girth weld and weld overlay/liner triple points were cut out. The girth welds and weld overlay/liner triple points were inspected visually and with dye penetrant inspection. Furthermore, at each strain gauge axis, measurements of pipe wall thickness on either side of each weld and the axial misalignment at the weld root were recorded. It should be highlighted that UT using a composite inspection tool was performed on the specimen (SP4), after completion of the test, to inspect the weld overlay/liner transition points. This technique is currently under development and was used on one specimen to generate data and help with the qualification of the tool. The inspection tool was first used on the specimen (Sp4) before testing to establish the baseline and was then performed on completion of the test (i.e., after reaching run-out life). However, no detectable crack indications were found during the inspection on completion of testing.

The calculated target lives for each specimen at the test stress ranges are summarized in Table 2. It should be noted that due to uneven stress distribution caused by residual ovalization during full-scale bending tests, the stress range values were conservatively calibrated to reflect the lowest measurement among all strain gauges.

Table 3 presents the performance of the triple points under cyclic loading conditions. Additionally, the fatigue performance of specimens Sp1, Sp2, Sp3, and Sp4 is graphically presented in Figs. 13 and 14 as S–N diagrams with double logarithmic axes. Figure 13 illustrates the fatigue performance of the transition points on the ID with DNV classes (D) and (E) design and target curves, whereas Fig. 14 shows the fatigue performance of the transition points on the OD with DNV classes (C) and (D) design and target curves.

Table 4 depicts the fatigue performance of the girth weld using the OD stress range. Figure 15 presents the fatigue performance of the girth welds on the OD with DNV classes (C) and (D) design and target curves.

Test specimen (Sp1): This specimen failed in the girth weld (GW5) at 10 o’clock (95 mm from the strain gauge G23) at 1,182,440 cycles without any sign of water leakage. The pipe exceeded the DNV class (E) target on the ID. The crack was approximately 190 mm long on the outside pipe surface. The crack was initiated by a weld repair defect at the weld cap toe of the girth weld (see Figs. 16–18 for crack illustration).

The failed girth weld was cut out and the four transition points and remaining girth weld root were inspected with DPI. No crack indications were found. After inspection, the specimen was repaired and the test was resumed. After weld repair, the test continued until failure at a total of 2,231,380 cycles in the drive end of the pipe at the 9 o’clock position at the girth weld (GW6) (near the strain gauges G7) again without any sign of water leakage. The specimen did not meet the target of twice the DNV class (E) target on the ID but exceeded the DNV class (C) target on the OD. The crack was 78 mm from the girth weld toes and was approximately 80 mm long on the outside pipe surface (see Figs. 17 and 18). Visual examination revealed evidence of surface flaws and grinding marks on the outer surface of the carbon steel at the crack location but also all around the pipe circumference (see Figs. 18 and 19). Furthermore, perturbations in strain gauge responses were observed before both failures.

Post-test inspection revealed that the crack initiated from flaws on the outer surface of the carbon steel pipe and propagated through the pipe wall thickness (carbon steel and clad overlay weld), see Figs. 17–19. Multiple crack initiations were observed along the carbon steel outer surface (see Figs. 20–24). No detectable crack indications were found during the DPI of the triple points and girth weld (GW05) at the end of the test.

Further metallographic examination of the transverse sections showed that the crack propagated in the clad overlay close to the transition point (Fig. 25). No cracking was observed at the triple point and weld overlay/liner transition point (Fig. 26). At the flaw location, a change in microstructure was observed (Fig. 27). Refined grains were observed around the crack initiation near the pipe's outer surface revealing that a heat event (probably due to the generated heat during the grinding activities) occurred at the flaw locations. Additionally, macro sections of the triple point at the transitions B1-R were taken around the pipe circumferences at 3, and 12 o’clock (see Figs. 28 and 29). For brevity, the micro sections at 3, and 6 o’clock were not presented in the paper but they are the same at 12 o’clock showing no cracks at the triple points.

It should be highlighted that the feature shown in Fig. 29 has characteristics that are compatible with a typically noted, low-range type of discontinuity whose origin is the welding of the triple point. As no growth was observed during FSBT/FSFT, no follow-up investigation was performed. This feature is a shrinkage-like flaw that often evolves with micro-cracks between the dendritic growth to a short range where the microstructure is enrichened by what DNV calls “deleterious phases.” Examples of the microsection undertaken for the triple point after completion of the full-scale testing are also shown in Figs. 30 and 31, which indicate that no cracks were observed in the triple point. The figures also demonstrate that after the full-scale bending followed by the full-scale fatigue, there is no visible growth on the post-mortem macros can be identified. This proves that there is no concern with the triple point if the triple point/tri-metallic joint follows high-quality manufacturing and welding procedures.

Test specimen (Sp2): This specimen was first stopped at 6,652,180 cycles (class E on the ID target number of cycles). No water leakage or change in the strain gauge reading, which would indicate cracking, was detected. The specimen was inspected by remote DPI. No crack indications were detected. The specimen was re-tested and was then stopped at 13,553,550 cycles after reaching the final target endurance (twice the DNV class target E). No indications of cracking were found during the DPI at the end of the test.

Test specimen (Sp3): This specimen was initially stopped at 1,102,820 cycles, corresponding to the DNV class (E) target number of cycles on the ID. During this phase, no signs of water leakage or changes in strain gauge readings were observed, indicating the absence of cracking. The specimen underwent DPI, and no crack indications were found. After this initial inspection, the specimen was returned to the testing machine to resume testing. The test continued until the final target endurance was reached at 2,519,150 cycles. At the end of the test, another DPI inspection confirmed no detectable cracks had formed. The results showed that the triple points of this specimen exceed the fatigue performance requirements for DNV class (E) on the ID, whereas the results for the OD stress range demonstrated compliance with the DNV class (D) target curve. This confirms that specimen Sp3 met the desired endurance targets with no failures.

Test specimen (Sp4): This specimen was initially stopped at 7,083,440 cycles, achieving the DNV class (E) target number of cycles on the ID. Similar to specimen Sp3, no signs of water leakage or changes in strain readings were detected, indicating no cracking. DPI was performed, and no crack indications were observed. The specimen was then returned to the testing machine to resume testing, reaching the final run-out life at 12,485,160 cycles. DPI inspections conducted at the end of the test confirmed no detectable cracks had formed. The results demonstrated that the fatigue performance for this specimen, with the triple points, exceeds the DNV class (E) target curve on the ID and meets the DNV class (D) target curve on the OD. The results also showed that all weld overlay/liner triple points met or exceeded fatigue performance expectations, except for girth weld GW-01, which did not meet the DNV class (C) target curve on the OD. Nevertheless, the endurance tests indicated excellent fatigue performance in the triple zone, with the joint at the triple point exceeding the class (C) target number of cycles on the OD.

Based on the testing carried out, the following facts about tested specimens can be drawn:

• Test specimens (Sp1 and Sp2): Both specimens and all triple points exceeded the DNV class (D) target curve on the ID and DNV class (C) target curve on the OD with no detectable liner cracking. All girth welds qualified to the DNV class (E) target curve on the ID and DNV class (D) and class (C) target curves on the OD, apart from the girth weld for GW-07 which only qualified to the DNV class (E) target curve on the ID.

• Test specimens (Sp3 and Sp4): All triple points and girth welds, apart from one girth weld (GW1), were located on the specimen (Sp4) test and were qualified for DNV class (D) target curve on the ID and class (C) on the OD. The girth weld (GW1) only qualifies for the DNV class (D) target curve on the OD. The results for both test specimens exceed the DNV class (E) target curve on the ID and the DNV class (D) target curve on the OD with no detectable liner cracking. The tests were stopped after reaching the target curves and were not run until failure. The selected target curves were chosen based on a survival probability of 97.7% and a minimum confidence level of 95%; the target curve was calculated following BS7608.

The results of the FSFT of CRA-MLPs offer crucial insights into the fatigue performance of the triple point. This junction, where the CRA liner, weld overlay, and backing steel pipe intersect, is prone to complex stress interactions that are not fully addressed by current industry standards, such as API-5LD, DNV-ST-F101, and DNV-RP-C203. By providing comprehensive data on the fatigue resistance of the triple point, this study fills a significant gap in these standards, emphasizing the need for specific fatigue benchmarks for high-stress regions.

This study aligns with previous research on the fatigue behavior of CRA-MLPs but advances the field by employing a full-scale testing approach. Unlike component-level studies, the full-scale perspective here offers a holistic understanding of how the entire CRA-MLP system, particularly the triple point, performs under operational conditions, which is essential for ensuring long-term pipeline integrity. The results indicate that, with proper manufacturing and installation, the triple point can exhibit strong fatigue resistance despite the lack of specific guidelines in current standards.

The findings of this study have significant implications for subsea pipeline design and maintenance. The observed fatigue resistance at the triple point suggests that industry standards and practices might need updates to better account for this critical area, potentially leading to improved pipeline safety, reduced maintenance costs, and enhanced reliability in subsea operations. Future studies could further explore the impact of various operational conditions (e.g., temperature, pressure) and a broader range of CRA materials to better understand material properties' influence on fatigue resistance and support new industry guidelines.

ECA at the triple point is challenging due to the complex geometry and the combination of materials, which complicates defect detection, stress evaluation, and the preparation of representative specimens for fracture toughness testing. While AUT and its advanced variants, such as PAUT and time-of-flight diffraction, are effective for girth weld inspections, their application at the triple point is hindered by the unique characteristics of CRA materials, such as signal attenuation and noise. RT, despite its limitations in measuring through-wall flaws, excels at detecting planar defects when aligned with the X-ray beam, particularly for cracks extending through the wall. This highlights the need for representative specimens to refine fracture toughness testing and validate inspection techniques. The development of a workmanship equivalent ECA could address these challenges by establishing practical defect tolerance criteria, ensuring both safety and performance in demanding applications​.

This study demonstrates the reliable performance of the triple point in CRA-MLPs under fatigue loading. All tested specimens exceeded the DNV class (D) target curve on the ID and the DNV class (C) target curve on the OD. The absence of crack growth at the triple point under simulated installation and operational stresses indicates that, with high-quality manufacturing and installation, CRA-MLPs are suitable for subsea applications under dynamic loads. These findings highlight gaps in current standards, such as API-5LD and DNV-ST-F101, particularly regarding the qualification and inspection of the triple point. To ensure the long-term integrity of submarine pipelines, future standards should incorporate specific performance criteria for this critical region. The resilience observed at the triple point reinforces the overall durability of CRA-MLP systems. While this study focused on the triple point, it raises the possibility that other high-stress areas, such as girth welds, may exhibit similar fatigue resistance under comparable conditions, provided they meet high-quality manufacturing standards. These insights support the design of robust, fatigue-resistant pipelines for subsea applications. Given these findings, industry standards would benefit from incorporating specific provisions for testing and inspection at the triple point. Establishing cyclic loading endurance requirements and acceptable crack growth rates as baseline criteria would enhance pipeline safety and reliability. Integrating these criteria into standards like DNV-ST-F101, DNV-RP-C203, and API-5LD could reduce reliance on customized ECAs, offering greater clarity and consistency for demanding submarine applications. The complexity of performing ECA at the triple point highlights the need for further research into tailored assessment techniques and advanced inspection methods. Overcoming these challenges would enable more precise defect detection, sizing, and evaluation, paving the way for industry standards that better address fatigue performance in complex geometries.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.