Steel catenary risers in sour deep water environments

an Engineering Critical Assessment

TWI Bulletin, March - April 2010

Steel catenary risers are used in deepwater oil and gas developments to transfer produced fluids from the seabed to surface facilities. SCRs can be subject to fatigue loading from a variety of sources including wave and tidal motion, vortex induced vibration (VIV) and operating loads. As Colum Holtam explains, when the production fluids are sour (ie contain water and H₂S), higher fatigue crack growth rates (FCGRs) are expected, and this can have a significant effect on defect tolerance.

An engineering critical assessment (ECA) is a fracture mechanics-based approach, used to evaluate the significance of a flaw, based on a particular combination of material, stress and environmental conditions. An ECA approach can therefore provide maximum allowable flaw sizes at the manufacture and installation stage to ensure that, for example, girth weld flaws do not reach a critical size during the projected life of the component. This differs from conventional fatigue design philosophy which uses a stress-life (S-N) approach, whereby an endurance curve is generated from a series of representative tests. S-N design curves can be found in BS 7608 (BSI, 1993) and DNV RP-C203 (DNV, 2005), for example, and will typically be based on the statistical mean of the experimental data minus two standard deviations of log N.

When using S-N design curves to evaluate offshore structures, a safety factor (typically up to 20) is used to ensure a conservative estimate of fatigue life. This factor of safety depends on, for example, the consequence of failure and confidence in the assumed fatigue loading spectrum (ie a high consequence of failure and an unpredictable loading spectrum would warrant a higher factor of safety). In a corrosive environment (eg sour service), a further fatigue life reduction factor (or knock-down factor) needs to be applied to account for the detrimental effect that the environment may have on fatigue performance. There are relatively few published S-N data for steels in a sour environment and as environmental factors such as pH, partial pressure of H₂S, and temperature can have a significant influence on the observed behaviour, it is necessary to generate project-specific S-N or crack growth rate data to assist in design.

The S-N approach can be used to assess the performance of a nominally defect-free weld, although joint misalignment can be allowed for. However, for welds with known defects, an ECA approach is required to demonstrate adequate fatigue life. While it remains appropriate to incorporate a safety factor (with respect to fatigue life) within ECA calculations, it is a matter of debate as to whether the factor used should be the same as that used in S-N calculations. In some cases it is believed that doing so would be overly conservative.

Ensuring safe riser design is the primary objective, but it is recognised that estimates of fatigue life that are overly conservative would result in greater numbers of repairs at the design and manufacture stage, lower operational lives and possibly more stringent inspection intervals. Each of these has the potential to make SCRs unviable from either a technical or an economic standpoint. While the safety factor with respect to fatigue loading can be a critical parameter when conducting an ECA, other inputs such as the assumed toughness, residual stress or fatigue crack growth law can similarly have a significant influence, particularly under sour service conditions.

The aim of this paper is to provide guidance for performing ECAs on internal surface-breaking defects in SCRs operating in a sour environment and subject to VIV fatigue loads. Example ECA calculations are presented for circumferential girth weld flaws in a typical SCR, based on the failure assessment diagram (FAD) approach within the framework of BS 7910 (BSI, 2005), using TWI's commercial software package CRACKWISE 4. The influence of certain key (mechanical) input variables is demonstrated, including the FCGR, determined from recent sour test results generated as part of this Core Research Programme (CRP) project. It should be noted that BS 7910 currently contains only limited guidance on the assessment of flaws in sour service conditions. Methods for the determination of the critical stress intensity factor for stress corrosion cracking (K_ISCC) and the assessment approach when using this type of laboratory data are still the subject of ongoing research.

Objective

Provide guidance for performing ECAs on internal surface-breaking defects in SCRs operating in a sour environment and subject to VIV fatigue loads.

Approach

Failure assessment diagram

To assess resistance to static failure, it is possible to represent material properties, defect geometry and loading conditions in a mathematical form, and generate what is known as an FAD. In this regard, the FAD approach can be considered as independent of component geometry. An FAD represents a two-parameter approach, acknowledging that either brittle or ductile fracture might occur, or plastic collapse. For fracture to occur, the stress intensity factor at the crack tip must be greater than the material's fracture toughness (or critical stress intensity factor). However, plastic collapse can also occur if the stress is high relative to the material proof stress or ultimate tensile strength (UTS), hence the need for a two-parameter approach to combine these two possible failure mechanisms. A generalised Level 2A FAD is shown in Figure 1, which is suitable for materials that do not exhibit a yield discontinuity (commonly referred to as Lüders plateau).

The vertical axis of the FAD represents the criteria for brittle or ductile fracture, often known as the fracture toughness ratio (K_r), which is the ratio of stress intensity factor (K_I) to material fracture toughness (K_mat). Both primary and secondary stresses (such as residual stresses) contribute to the applied stress intensity factor. The horizontal axis represents the likelihood of plastic collapse, often known as the load ratio (L_r), which is the ratio of the in-service (or reference) stress to yield strength (or 0.2% proof strength) or flow stress. Secondary stresses (such as residual stresses) do not contribute towards plastic collapse. The fracture toughness and load ratios are connected via a locus, and the proximity to this locus is an indication of how close a specific material, flaw and stress combination is to failure.

As indicated in Figure 1, on or below the locus represents an acceptable or stable flaw, whereas a point outside the locus represents an unacceptable flaw where the risk of failure cannot be tolerated. If the damage mechanism and service conditions are adequately understood, then predictions can be made as to how a flaw might grow and what size flaw can ultimately be tolerated in-service, assuming no interaction with adjacent flaws.

For internal surface-breaking flaws in a sour environment, the dominant form of crack extension can be sulphide stress cracking (SSC), which can occur at applied stress intensities much lower than the material's fracture toughness. At applied stress intensities higher than a critical value (K_ISCC) flaws will grow, usually rapidly, until their size reaches a point where fracture or plastic collapse intervenes. It is therefore necessary to restrict the applied stress intensity factor to a value less than K_ISCC. It is then possible to account for sour service conditions by using K_ISCC, generated from laboratory data, as an estimate of the material's fracture toughness in a sour environment and therefore define the FAD envelope in terms of sour service properties. This differs slightly from the approach defined in BS 7910, but is believed to be conservative.

Flaws considered

The case considered in this report is a circumferential surface-breaking flaw on the internal surface of an SCR, located at the root of a girth weld (see Fig.2). The relevant BS 7910 stress intensity factor and reference stress solutions for this type of flaw are M.4.3.3.2 and P.4.3.2 respectively (BSI, 2005).

The case of an embedded or external flaw has not been considered in the current work. An external flaw would be exposed to a seawater environment, which is known to be less severe than a sour environment in terms of increased FCGR. The case of an embedded flaw is more complicated as the flaw could propagate through material that contains a significant amount of absorbed hydrogen, even though the environment does not have direct access to the crack tip region. Use of a crack growth law derived from tests conducted in a sour environment is probably conservative in this instance, but perhaps overly so. An experimental study to investigate this aspect of material behaviour is underway at TWI.

For assessment purposes, surface-breaking flaws are idealised as being semi-elliptical. Standard practice is to calculate an applied stress intensity factor, K, at the base of the flaw (ie the deepest point) and at the surface of the flaw. For fatigue assessments the corresponding increments of crack growth at both locations are calculated, while for static failure assessments the maximum (or worst case) is used.

Input parameters

SCR geometry and initial flaw dimensions

The SCR is assumed to be constructed from C-Mn X65 seamless API 5L parent material, with dimensions as indicated in Table 1. An aspect ratio of 0.1 (flaw height to length) was chosen for the initial flaw size.

Table 1 Assumed geometry of SCR

Weld geometry

The weld is assumed to be a full penetration girth weld produced using mechanised processes. A typical example is shown in Figure 3. Weldments, particularly the heat affected zone (HAZ), can be susceptible to SSC due to the hardened material (untempered or partially tempered martensite) that may be present in this region. Welds designated for sour service are restricted to a maximum hardness limit of 250HV at the weld root. The internal surface-breaking flaw is therefore assumed to be located at a girth weld, close to the weld root toe.

Stress concentration factor due to presence of weld (M_k factor)

The local stress concentration at the weld toe is characterised using the parameter M_k. Standard solutions for surface-breaking flaws are provided in Annex M of BS 7910 (BSI, 2005), derived from 2D (and for certain geometries 3D) finite element analyses. M_k is dependent on what is termed the attachment length, which in the case of an internal surface-breaking flaw in a full penetration pipeline girth weld is the width of the weld root protrusion. For the purposes of this assessment, a weld root width of 4mm was adopted (Table 1). M_k is maximum near the surface of the pipeline and its influence decreases as flaw depth increases. M_k is calculated automatically within CRACKWISE 4 for the selected attachment length and in this assessment the standard 2D solutions for an internal surface-breaking flaw were used (M.4.3.3.2) (BSI, 2005), assuming a full penetration weld.

Axial misalignment

Axial misalignment can be defined as the offset between the centrelines of the pipe wall across the girth weld. Axial loading of a misaligned joint generates a bending stress at the location of the flaw. In the current work a maximum level of misalignment of 1mm was assumed. The stress magnification factor, K_m, was calculated, based on this assumed misalignment, using Equation 3.3.3 from DNV-RP-C203 (DNV, 2005), which is known to be slightly less conservative than the equivalent equation in BS 7910 and is included in CRACKWISE 4.

where ∂ is the misalignment, T and t are the wall thicknesses either side of the weld, D is the outside diameter of the pipe, L is the attachment length (or width of the girth weld cap) and;

A weld cap width of 12mm was assumed (Table 1). This resulted in K_m = 1.13.

Welding residual stresses

Welding residual stresses are dependent, amongst other things, on the welding process used and whether any stress relief was performed. SCR girth welds tend not to be heat treated and therefore a uniform welding residual stress of tensile yield strength magnitude was assumed in the calculations, in line with the guidance provided in BS 7910 (7.2.4.1) (BSI, 2005). This stress will relax under the influence of applied load, but not significantly if the applied loads are low. The effect of lower residual stresses was also investigated.

For an internal circumferential surface-breaking flaw at a pipeline girth weld, the residual stress profile perpendicular to the flaw (and in this case the weld) will be limiting. While the assumption of uniform residual stresses of yield strength magnitude is a common approach, it is also recognised as being conservative, even if the residual stresses are allowed to relax under applied load. For relatively thick walled pipe it has also been shown that residual stresses tend to be lower or even compressive on the inner surface of a pipeline girth weld but it is difficult to provide quantitative guidance. A combination of residual stress measurement and modelling (for a specific welding procedure) does however provide a possible method for determining an appropriate but less conservative assumption regarding the assumed residual stress state.

Tensile properties

The tensile properties used in this report (Table 2) are based on tests carried out at TWI, using the same X65 parent material as that used to generate the FCGR data.

Table 2 Room temperature material properties used in the assessment

Static stresses

A maximum axial static stress of 150MPa has been assumed in the current work, representing a typical maximum operating stress for an SCR.

Fatigue stresses

Fatigue stresses can result from numerous sources including the pressure cycles experienced by the riser during operation, wave and tidal motion and VIV. The extent of fatigue damage is dependent on the combination of stress range and the number of cycles. SCRs can experience very high numbers of low stress cycles due to VIV, a phenomenon caused by the constant passage of marine currents past a riser, causing turbulence.

Experience from the offshore industry suggests that the majority of fatigue damage that occurs in SCRs is due to relatively low stress range cycles (ie due to the high number of cycles at these low stresses and the relatively few cycles at higher stresses). In this regard, stresses due to VIV often tend to dominate for the specific case of SCRs. Shutdown-restart sequences, which can dominate the fatigue assessment of high pressure, high temperature flowlines subject to lateral buckling, tend not to be significant in the case of SCRs. In the current work, the life of the SCR is also not affected by the position of the flaw around the circumference of the pipe, as is the case with the assessment of flaws in reeled pipelines for example. For stresses due to VIV, a safety factor of 20 is often applied to design life to provide a conservative prediction of fatigue life.

The first weld at the top of the SCR and the touchdown point (TDP), where the nominally vertical riser meets the pipeline or flowline on the seabed, are the most stressed locations and therefore critical in terms of fatigue loading. The critical location for VIV is often the top of the riser since this is subject to higher stress ranges than the TDP, if only for a small number of cycles.

The fatigue spectrum used in the current work is based on a design life of 30 years for the SCR. A typical VIV fatigue spectrum for the TDP and top weld of an SCR was scaled such that a simple S-N analysis (using a Class E design curve and a knock-down factor of 30 for the sour environment) gave a target design life of 600 years (ie 30 years with safety factor of 20). For stress ranges below the constant amplitude fatigue limit, a reduced fatigue slope was used in this S-N analysis, as recommended in BS 7608 Section 4.4 (BSI, 1993). The final annual fatigue spectrum for both the TDP and the top weld are shown in Figure 4. It can be seen that the TDP is subject to a greater number of low stress cycles whereas the top weld has a greater number of higher stress cycles but fewer low stress cycles.

ECA safety factors

As noted above a safety factor of 20 has been assumed in the S-N analysis, and this has been used to scale the fatigue loading spectrum accordingly. It is however standard practice to apply a smaller safety factor on fatigue life when performing ECA calculations. For example a safety factor of 10 would mean a target life of 300 years, a safety factor of five would mean a target life of 150 years and no safety factor at all would reduce the target life to 30 years. All three options were investigated in the present analysis.

Fatigue crack growth rate laws

The assumed FCGR usually takes the form of the Paris law which relates the crack growth per cycle to the stress intensity factor range (ΔK), as shown in the equation below, where m and C are constants.

However, below a certain value of ΔK (known as the threshold), no fatigue crack growth is expected. BS 7910 contains standard crack growth laws for steels in air and in seawater environments. However, in a sour environment, FCGRs can be significantly higher and so an appropriate crack growth law needs to be determined from laboratory testing.

Representative data generated from a decreasing ΔK test carried out using a stress ratio of 0.5 in an aqueous solution of 5% sodium chloride and 0.4% sodium acetate, acidified to a pH of approximately 3.5 and saturated with a mixture of 7%H₂S in N₂ are shown in Figure 5. The test was carried out at 25°C (±3°C). Also plotted is the mean curve for steels in air (R ≥ 0.5) taken from BS 7910, and the design curve for steels in air (ie mean plus two standard deviations of log da/dN).

At high ΔK the FCGR is approximately 30 times higher than the mean curve for steels in air. An appropriate crack growth law for a sour environment can therefore be determined by offsetting the design curve (mean plus two standard deviations) for steels in air by a similar amount. As illustrated in Figure 5, this provides a relatively simple means of determining an appropriate upper bound to ensure a conservative assessment, but does not take advantage of the fact that the influence of the sour environment appears to be less significant at lower ΔK.

It can be seen that at lower ΔK (400-300Nmm^-3/2), the FCGR decreased rapidly (more than an order of magnitude) and approached the design curve for steels in air. The reduced influence of environment at low ΔK has also been reported elsewhere.

It is likely that adopting a crack growth law which provides a better fit to the experimental data at low ΔK will have a significant influence on associated ECA calculations. A four stage law (see Fig.5) has therefore also been used in the current work to assess the extent to which this affects the determined fatigue lives.

Fracture toughness and K_ISCC

Values of K_ISCC are known to depend on environmental parameters such as pH, partial pressure of H₂S and temperature. At ambient temperature, and low pH (around 3.5), K_ISCC may be in the range 800-1600Nmm^-3/2 (25-50MPam^0.5). However, in less aggressive service environments (eg pH5 and elevated service temperature), values of K_ISCC may be substantially higher. A K_ISCC value of 3160Nmm^-3/2 (100MPam^0.5) was assumed in the current work, based on TWI's project experience.

End of part I. Part II, in the next edition of Bulletin reveals the results of Colum Holtam's work and includes assessments, conclusions and recommendations drawn from his extensive investigations into the behaviour of steel catenary risers in sour deepwater environments.