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

IPC2012-90620

FULL SCALE EXPERIMENTAL DATABASE OF DENT AND GOUGE DEFECTS TO IMPROVE BURST AND FATIGUE STRENGTH MODELS OF PIPELINES
Mures Zarea GDF SUEZ R&I Division - CRIGEN St Denis La Plaine, France Brian Leis Battelle Memorial Institute Columbus, Ohio, USA Geoff Vignal Enbridge Pipelines Edmonton, Canada 1. ABSTRACT External interference on gas and oil transmission pipelines is consistently reported as leading cause of leaks in Europe and USA as identified in the EGIG and PHMSA incident databases. External interference due to ground working machinery strikes, rock strikes during backfilling, etc. on buried pipelines result mainly in dent and gouge defects. The long-term integrity of a pipeline segment damaged by a dent and gouge defect is a complex function of a variety of parameters, including pipe material properties, pipe geometry, defect geometry linked to indenter shape, aggression conditions. The complexity and extreme variability of these dent and gouge defect shapes and pipe materials lead simple assessment models to scattered predictions, hinting towards an insufficient description of real structural and material behavior. To improve knowledge beyond the numerous studies led in the past, and to provide a sound foundation for developing and validating mechanistic models for predicting burst and fatigue strength of such defects, a large experimental program was funded by PRCI and US DoT and further coordinated with a complementary EPRG program. The experimental program part dealing with combined Dent and Gouge defects is covered for modern pipes (24 OD, X52 and X70) by PRCI project MD-4-1: realistically created (with a Pipe Aggression Rig) defects submitted to full scale burst and fatigue tests, in addition to extensive characterization. This work interfaces with modeling to predict the immediate burst strength and fatigue resistance of such damage in two PRCI projects denoted MD-4-3 and MD-4-4 respectively. This paper gives an overview of some of these activities: PRCI project MD-4-1 results: material characterization, full scale burst and fatigue tests on Dents with Gouges, as well as detailed explanations concerning the initial approach to model burst and fatigue life of these defects, as developed byr PRCI project MD-4-4. The final outcome of the expected knowledge improvements about the mechanical strength of dent and gouge combinations will be applicable by pipeline operators, in order to enhance integrity management systems designed to manage the threat associated with mechanical damage. 2. INTRODUCTION Mechanical Damage on gas and oil transmission pipelines is consistently reported as a leading cause of leaks on pipelines around the world as identified in the EGIG [1] and PHMSA [2] incident databases. Mechanical damage defects result from direct contact of an indenter with the pipe and are due to a number of causes; the pipe resting on rock, ground working machinery strike, rock strikes during backfilling, amongst others. The severity of damage incurred to the pipeline varies significantly depending on mode of cause. During the construction sequence, minor damage such as shallow scraping or scoring of the outer pipe surface can occur in dents when the pipe settles onto a rock and is manipulated into position. Conversely, severe damage such as sharp dents containing deep gouges can result from dynamic impacts from construction equipment such as backhoes. The two mechanisms for inducing mechanical damage described above demonstrate opposite ends of the spectrum in Remi Batisse GDF SUEZ R&I Division - CRIGEN St Denis La Plaine, France Philippe Cardin GDF SUEZ R&I Division - CRIGEN St Denis La Plaine, France

Copyright 2012 by ASME

terms of severity of damage, however, both forms of damage have been known to cause pipeline failures. The rock induced damage typically results in delayed failure, while the impact from construction equipment has a higher potential for immediate failure. Managing the threat of mechanical damage to pipelines relies of a pipeline operators ability to detect, identify, locate, characterize and mitigate the damage through integrity management programs. Understanding the behavior of mechanical damage and represent it by mechanistic models can support development of guidelines for maintenance strategies including appropriate timing for repair and indications for appropriate repair options. The continued development of mechanical damage assessment models is a priority also to support decision making during the assessment of damage features, prioritization of damage features for further characterization / repair and the safe execution of pipeline repair activities belonging to pipeline integrity programs. Understanding the limitations and practical application of mechanical damage assessment models is also of importance, and is addressed in the discussion presented in this paper. We focus here on the leading cause associated with pipeline failures, i.e. external interference, producing in general dents associated with gouges. The long-term integrity of a dented and gouged pipeline segment is a complex function of many parameters, including pipe material properties, pipe geometry, defect geometry, indenter shape, trajectory & velocity, and pressure history at and following indentation. In order to estimate the safe remaining operational life of a pipeline with dent and gouge defects, all of these factors must be accounted for in the analysis. The complexity and extreme variability of dent and gouge defect configurations lead simple assessment models to uncertain predictions, hinting towards an insufficient description of real structural and material behavior. In order to improve knowledge beyond the numerous studies led in the past, and to provide a sounder foundation for developing and validating mechanistic models for predicting burst and fatigue strength of such defects, a large experimental program was funded by PRCI and US DoT. This research program collects more detailed and comprehensive experimental data to facilitate the improvement of mechanical damage assessment models for dents with other features: gouges, metal loss and welds. While the full experimental tests program consists in two related PRCI projects (MD-4-1 and MD-4-2) that share the material characterization work on modern pipe steel grades, X52 and X70, and in the DoT project # DTPH56-08-T-000011 (the DoT project) aimed at testing vintage X52 grade, we focus here only on MD-4-1 that focuses on dent and gouge defects.

MD-4-1 (modern pipe), its extension MD-4-6, and the DoT #339 project (latter two on vintage pipes) share the same approach for Dent and Gouge damage: creation of three similar samples of realistic Dent and Gouge defects with different sizes of combined dent and gouge. Realistic defect creation means the pipe is hit with an excavator-like Pipe Aggression Rig fitted with an excavator tooth so that gouge and dent are created simultaneously. Two samples are used for burst and fatigue tests with very detailed instrumentation. After failure, a thorough failure investigation highlights the failure mechanisms and the issues of crack initiation and propagation. The third sample is used for detailed characterization, in terms of geometry, of residual stresses, presence or absence of microcracks, etc. Tests are performed by GDF SUEZ Research and Innovation Division, CRIGEN, France. These defects have extreme yet realistic dimensions and features as found in the field: shallow dent and intermediate gouge, deep dent and intermediate gouge. This work interfaces with modeling to predict the immediate burst strength and fatigue resistance of such damage in two PRCI projects denoted MD-4-3 (burst strength) and MD-4-4 (fatigue strength with aspects of burst strength embedded) respectively. Prediction results from a first approach developed in MD-4-4 are also reported and discussed here. Main result examples reported here are: This paper shows the test matrix, test conditions and representative results from burst and fatigue tests on dent and gouge defects on modern steel pipes from the MD-4-1 project. On the modeling side, it details ongoing efforts to develop improved pipeline dent and gouge fatigue integrity assessment models underway in PRCI project MD-4-4. Modeling of fatigue failure is performed by Battelle Memorial Institute, Ohio, USA. It should also be mentioned that this effort is coordinated with EPRG, that manages two complementary projects on burst strength of dents with gouges. 3. MATERIAL CHARACTERIZATION Two different pipes are currently being used in the full scale test program and are identified as Pipes A and B. Pipe A and B are modern materials of Grades X-52 and X-70 respectively. Their outer diameter is 609.6 mm diameter. The nominal thicknesses of Pipes A and B C are 7.9 mm and 8.89 mm respectively. Diameter versus wall thickness ratios (D/t) for Pipe A and B are 77 and 69, respectively. All the joints from pipes A and B came from the same heats. Pipe A and Pipe B have been used in both the MD-4-1 and 4-2 programs. A vintage pipe will be used in the MD-4-1 program to start extending coverage of results also to existing steel grades. 3.1 Baseline Material Characterization Detailed material characterization tests were carried out on the two pipes and included chemical analysis, microstructure evaluation, tensile tests, Charpy impact testing and CTOD

Copyright 2012 by ASME

testing and are detailed in Tables 2.1 - 2.5 respectively. These are modern, clean, strong and tough steels. Composition (wt%) C Mn S P Si Cu Ni Cr V Cb Sn Mo Al Ca B Ti N Pipe A 0.04 0.83 <0.005 0.008 0.23 0.21 0.13 0.07 0.006 0.023 0.008 0.054 0.038 <0.005 <0.002 <0.005 0.013 Pipe B 0.04 1.60 0.007 0.007 0.16 0.27 0.13 0.07 <0.005 0.071 0.008 0.198 0.049 <0.005 <0.002 0.020 0.010

distinctly hardened outer layer and a strain hardened layer beneath it were created at the gouge surface as shown in Figure 3.2.1.

Base Metal

Strain-hardened Base Metal Layer

Hardened Layer with Tooth Metal

Figure 3.2.1 : Hardened layer at gouge surface of defect 2.1.3 The outer layer is most often between 50m and 100m thick and its hardness is very high, around 600 HV0.5. Initially, it had been assumed that the pipe steel turned into a martensitic phase due to the sudden thermal cycle due to the tooth impact very quick friction heating followed by rapid conductive cooling in a thin layer. Such high hardness values were not expected, so to check the initial assumption, additional investigations were carried out: several thermal tests were performed with a Gleeble furnace to reproduce this martensitic microstructure of the pipe steel. But it was not possible to reproduce such hardening, and the highest hardness achieved was around 275 HV, far below 600 HV. Alternatively, the Blondeau formula was used giving the martensitic hardness versus the chemical steel composition and cooling rate tr between 800C and 500C : HV=127+949 *%C+27 *%Si+11*%Mn+8*%Ni+16*%Cr+21*log(tr) Using the pipe steel chemical composition and estimating conservative cooling rates between 1C/s and 5C/s, the calculated hardness results are between 200HV and 220HV. The reason is that the carbon content is too low to produce a hard martensitic microstructure. This calculation confirms the fact that the hardened layer is not pure pipe steel. In order to investigate further, a tooth used for impact tests was analyzed (Figure 3.2.2). Chemical and hardness analysis were performed on tooth samples and compared to the hardened layer characteristics. Hardness values are close (535 HV for the tooth and around 600 HV for the hardened layer). The chemical contents in table 3 show comparable values for the tooth and the hardened layer and differences with pipe steel concerning Silicium, Chromium

Table 2.1: Chemical Analysis Results for the two Pipes Pipe ID A B Inclusion Volume Fraction (%) 0.05 0.07 Pearlite Volume Fraction (%) 3.2 3.1 Ferrite Grain Size (m) 6.2 3.2

Table 2.2: Quantitative Metallographic Test Results for the two Pipes Pipe ID A B Orientation Transverse Longitudinal Transverse Longitudinal 0.5% Yield Strength (MPa) 437 450 506 546 Ultimate Strength (MPa) 516 503 667 652 % Elongation 35 34 23 30

Table 2.3: Pipe Body Tensile Strip Test Data for the two Pipes Additional investigation observed on defect 2.1.3 of the hardened layer

Pipe Specimen Upper Shelf Impact Energy, ID Size Joules (Average) A 2/3 160 B 2/3 87 Table 2.4: Base Metal Charpy V-Notch Impact Test Results of the two Pipes It was systematically observed that after dynamic impacts, a

Copyright 2012 by ASME

and Manganese. A slight kinematic effect on mechanical behavior is observed with an elastic area increasing in tension and decreasing in compression with increasing strain. Up to a 8.5% pre-strain, the high toughness level (Jc = 929kJ/m without pre-strain) is not significantly affected (Jc=668 kJ/m with 8.5% pre-strain) for this highly ductile modern material. The same behavior was found for the modern X70 material, therefore the influence of pre-strain on toughness should be more significant in less tough and clean steels. Samples for chemical and hardness tests Figure 3.2.2: Samples taken from a tooth for chemical and hardness analysis %Si 0.63 0.64 0.25 %Cr 0.63 0.63 0.09 %Mn 1.54 1.49 2.2 4 TEST ON DENTS WITH GOUGES Fifteen realistic combined Dent and Gouge defects were created with the Pipe Aggression Rig (Fig. 4.1.a) in modern pipes X52 and X70 in the frame of the MD-4-1 PRCI project and twelve more are planned to be created on vintage pipes in the DoT and MD-4-6 projects. The sizes of the defects were derived from an interpretation of defect dimensions found in the field, by favoring bracketing configurations, i.e. extreme conditions, either highly dynamic impact or slower dynamic impact, resulting in different types of defects. 4.1 Tests matrix and global results The matrix in Table 4.1 describes the experimental study. Defects are identified by three characters x.y.z : x identifies the pipe number; y identifies the type of combined defect (shallow dent and moderate gouge, severe gouge, severe dent and moderate gouge, etc.; z is the defect number for each type (1 for destructive characterization, 2 for burst test, 3 for fatigue test). Currently the MD-4-1 project on modern pipes is in its finalization stages, whereas the work on vintage pipes (MD-4-6 and DoT projects) is in its early phases. Each mechanical strength test is particularly well instrumented: During defect creation in highly dynamic or slower dynamic conditions with realistic indenter impact (tooth from excavator), forces and displacements are recorded in horizontal and vertical directions to evaluate defect creation energy versus time. Strain gauges are stuck on the internal pipe wall to record strains during indenter impact. For fatigue and burst tests, a wide range of instruments are used, like strain gauges (rosettes) around the gouges, opening displacement gauge: specifically for fatigue tests: LVDT, potential drop sensor, pictures of targets recorded with a camera synchronized with the pressure to record defect evolution with fatigue cycles, and a specific profiling tool to record the evolution of the longitudinal dent profile during the monotonic pressure increase.

Hardened layer Tooth Pipe steel

Table 3 : Comparison of chemical contents between hardened layer, tooth and pipe steel These results open the way to the interpretation of the creation of the hardened layer from the tooth material. But this interpretation has to be confirmed by a relevant detailed description of the deposition mechanism due to friction mechanical and thermal effects between the tooth and pipe steel during the impact. In summary, dynamic impact creates a very hard thin outer layer, and a strain hardened layer below, both around 50 100 m deep. Metallurgical investigation after defect creation indicates that micro-cracking is associated with the outer layer. 3.2 Cyclic behavior and influence of pre-strain on toughness Dent creation leads to local inverse pipe bending. In this area, the load history is not monotonic with local plastic compression or tension strains occurring during dent creation and then reversing to the opposite sign during re-rounding due to internal pressure after removing the indenter. So there is a load history dependence of flow properties [3] associated with mechanical dent damage which could be taken into account by a material kinematic hardening law. In addition, the local plastic pre-strain induced by dents may significantly affect material toughness [3 - 6]. For this reason, cyclic behavior tests and toughness tests - JCurves with pre-strain - were performed on the modern X52 and X70 materials. Figure 2.6 presents examples of results of the cyclic behavior and of the effect of pre-strain on modern X52 material.

Copyright 2012 by ASME

Figure 2.6.a: Engineering stress-strain cyclic behavior in modern X52 material.

Figure 2.6.b: Strain effect up to 8.5% on toughness in modern X52 material.

Copyright 2012 by ASME

Figure 4.2: Evolution of dent profile versus pressure for defect 1.3.2 and identification of bulging above 30 bar, defect creation pressure.

Figure 4.4.a: Evolution of targets displacement (black and grey symbols).

Copyright 2012 by ASME

Pipe number Type of defect Destructive characterization Burst test Fatigue test

MD-4-1 Project Pipe 1 Modern X52 Pipe 2 Modern X70 1 2 3 1 2 1.1.1 1.2.1 1.3.1 2.1.1 2.2.1 1.1.2 1.2.2 1.3.2 2.1.2 2.2.2 1.1.3 1.2.3 1.3.3 2.1.3 2.2.3

MD-4-6 Project Pipe 3DG Vintage 1 2 3.1.1 3.2.2 3.1.2 3.2.2 3.1.3 3.2.3

DoT#339 Pipe 4 Vintage 1 2 4.1.1 4.2.1 4.1.2 4.2.2 4.1.3 4.2.3

Table 4.1: Test matrix for combined Dent and Gouge defects MD-4-1, DoT#339 and MD-4-6. After defects creation, 3D laser mapping is performed to record the defect geometry and cracks are searched for by M.P.I. (Magnetic Particles Inspection). On defect x.y.1 residual stresses in the dent are determined by X Rays at the surface and in several locations along the cross-section by neutron diffraction (NIST in USA and Chalk River Reactor Facility in Canada are kindly contributing these very involved neutron diffraction measurements). Metallurgical and fracture investigation are performed after creation of defect x.y.1 and also after burst and fatigue tests to examine potential change of microstructure, presence of micro-cracks and to estimate experimentally levels of local strains in the dented and gouged areas.
Defect Tooth Aggression Energy (J) Pressure (bar) Dent depth (%) Gouge depth (%) Gouge length (mm)

Fatigue failure results (Table 4.3.b) exhibit a one order of magnitude range of number of cycles to failure: 2,007 to 20,494. One interesting observation is that a fatigue test in the pressure range of the dent bulging effect (see 4.2) leads to rapid failure as a function of number of cycles, even if fatigue pressure levels are low compared to burst pressure. Defect 2.2.3 loaded in the pressure range 20 bar - 60 bar in the dent bulging window (created at 20 bar) fails ten times more rapidly than defect 1.3.3 created at 30 bar and loaded with the same 40 bar pressure range, but well above the dent bulging window, between 53 bar 93 bar.
Defect Failure Pressure (bar) Theorical failure of pipe body (Barlows formula) Comments

1.1.1b 1.1.2 1.1.3 1.2.1b 1.2.2 1.2.3 1.3.1 1.3.2 1.3.3 2.1.1 2.1.2 2.1.3 2.2.1 2.2.2 2.2.3
C481 Cal44m C481 Esco Cal44

3424

85 85 85 85 85 85 30 30 30 85 85 85 20 20 20

1.1 1.6 1.3 2.6 2.6 2.5 5.3 5.9 6.1 1.5 1.6 1.5 4.7 5.2 5.3

26.5 6.3 11.3 43.0 34.2 46.8 27.8 29.1 20.2 22.2 18.9 20.0 20.0 16.7 21.1

135 150 135 110 115 105 331 375 321 175 200 165 331 353 319

1.1.2 1.2.2 1.3.2 2.1.2 2.2.2

133.3 110.3 130.9 185.1 193.5

132.7 146.0
(UTS=512 MPa-563.2MPa)

Rupture outside of defect in pipe body (not in seamweld ERW) Rupture in defect Rupture in defect Rupture in defect Rupture in defect

D D SD D SD

4713 5816 7676 6145 34930 28312 23973 7331 5912 4907 33974 26726 28412

132.7-146.0 132.7-146.0
(UTS=628 MPa-691 MPa)

185.4 204.0 185.4 - 204.0

Table 4.3.a: Results for Burst tests

1.1.3 1.2.3 1.3.3 2.1.3 2.2.3

45 bar 85 bar 45 bar 85 bar

10869 5200 20494 17700 2007

Several interruptions of fatigue test

53 bar 93 bar
(0.46 0.80 YS)

Pressure loading above the window of bulging effect Pressure max at 0.80 of current YS Pressure loading in the window of bulging effect

88 bar 128 bar

Table 4.2: Characteristics of dent and gouge defects created in MD-4-1. SD means Slower Dynamic aggression with a worn tooth and D Highly Dynamic aggression with a sharp new tooth. Defect created by SD is a long deep dent hence associated with higher defect creation energy. Table 4.2 summarizes all defect dimensions (residual dent depth at no pressure, gouge depth and length), and other characteristics, like internal pressure during pipe aggression, at 72% of actual yield pressure, except for the ~ 5%deep dents that could be created only with a lower pressure, worn teeth and a slower aggression speed. Table 4.3.a shows that the burst pressure of these damaged pipes is very close to the burst pressure of flawless pipes due to the high ductility of these modern pipes, the only exception being defect 1.2.2 which is a very severe gouge (depth 34%) in a shallow dent (2.6% deep).

20 bar 60 bar
(0.12 0.37 YS)

Table 4.3.b: Results for Fatigue tests Metallurgical post-failure investigation shows that the dent creation by dynamic impact with new sharp excavator teeth induced severe damage in materials with presence of hard layers associated with micro-cracks at the gouge surface (Figure 4.1.a and 4.1.b). The following sub-chapters provide more insight on one hand on instrumentation to record the longitudinal dent profile evolution during the pressure increase to burst failure, and on the other hand, on the gouge opening evolution during the fatigue test with observation of the crack-initiation and leak at the gouge surface via camera monitoring of targets during the fatigue test.

Copyright 2012 by ASME

evolution of the targets displacements follows that of the clip gauge opening displacement in a similar trend, with a larger amplitude. Slope changes are also more visible, hinting for initiation around 1200 cycles in this case. Crack-initiation (left) and leak (right) at the gouge surface Figure 4.1: Presence at the gouge surface of defect 1.1.1b.of a) cracks and b) hard layers (tooth metal at the surface, and hardened pipe body metal beneath 4.2 Example of instrumented burst test and results The vertical displacement along the dent at each time during the pressure increase measured with a continuous dent profiler provides the pressure range for the bulging effect Figure 4.2, defect 1.3.2), defined by the largest displacement for a given pressure increase. The dent starts to spring back just above its creation pressure, in this case 30 bar, with large displacements (up to 15 mm over 10 bar). This detailed knowledge of the dent re-rounding behavior is a completely original feature of this study, and was used to set the pressure range for the fatigue tests 1.3.3 and 2.2.3, as discussed above: in one case, cycling in the dent rerounding range (short fatigue life due to larger associated vertical displacements, see Fig. 4.3), in the other case, cycling well above the re-rounding range (long fatigue life, as vertical displacements are more limited, up to 2-3 mm for 10 bar, see Fig. 4.3). 4.3 Example of instrumented fatigue test and results To determine the gouge opening evolution during the fatigue cycles and to capture more precisely the instants of crackinitiation and leak appearance, a camera associated with targets on each side of the gouge were installed (Figure 4.3). Pictures are taken at each maximum and minimum cycle pressures.

Figure 4.4.b: Photos of crack-initiation and leak at the gouge surface. 5 CURRENT APPROACH IN MODELING FOR DENT AND GOUGE DEFECTS 5.1 Background for modeling approaches In complement to the full-scale experimental work just detailed, PRCI initiated a pair of independent modeling projects whose objectives were to assess 1) the immediate failure pressure of a dent and gouge defect (MD-4-3), and 2) its potential for possible pressure-cycle induced fatigue crack growth (FCG) at damage that did not lead to immediate failure (MD-4-4). As yet the details of Project MD-4-3 have not been released such that predictions of burst strength and critical defect sizes needed as inputs to the fatigue growth model are provided based on prior developments that had shown success in predicting damage severity, which is also a part of project PRCI PR3-9305 [10]. The approach for MD-4-4 was founded on the realization that if the dent and/or gouge, or local changes in steel properties diminish to zero, as could occur in a minor contact event, then the resulting circumstances are identical to the simple longquantified case that involves an isolated crack in a damage-free pipe (e.g. [11], [12]). Because this approach adapts already proven technology for fracture and FCG in undamaged pipes, it is technically sound as well as fiscally prudent. Conceptually, this mechanics/fracture framework quantifies the role of the dent and/or gouge in modifying the local stressstrain field, and accounts for other local changes like pre-strain on the properties to the extent required. Because of this basis in mechanics and fracture theory, MD-4-4 could use this existing technology independently of MD-4-3 to first predict combinations of failure pressure and damage size in terms of initial and final crack sizes, and then complete the fatigue predictions.

Figure 4.3: Target pairs 1-2 and 3-4 to follow the gouge opening displacement. Figure 4.4.a provides an evolution graph of targets displacement for defect 2.2.3, and Fig. 4.4.b photos when the crack initiation and leak appear at the gouge bottom. The

Copyright 2012 by ASME

5.2 Immediate failure predictions a first approach using existing information The basic approach to predict flow and fracture behavior, as well as fatigue crack growth (FCG) at cracks in pipelines traces to the early 1970s. In that era, major investments were made in offshore structures for oil production, [13] and in nuclear energy (vessels and piping, e.g. [14]). Aspects of the structural and life-assessment technology developed then remain largely viable today for applications adequately quantified in terms of linear elastic fracture mechanics (LEFM), and its crack-driving force known as the stress-intensity factor, denoted K [15], with only minor differences emerging since. In contrast, major developments have emerged since then to address scenarios requiring nonlinear [elastic plastic, EP] fracture mechanics [16] (NLFM), and the use of J-based crack-driving force, as adapted for example to pipeline applications (e.g., [12]). Fundamental to predicting failure due to cracking (e.g.,[14]) whether immediate or delayed is the balance between the motivation for cracking, and the materials inherent resistance to that cracking. For the case of an axial crack in a pipe, which exists as the lower-bound limit of dent and gouge damage, the crack driving force for applications where LEFM is valid has been expressed using a strip-yield plastic-zone [16]. For larger-scale crack-tip plasticity, nonlinear fracture mechanics (NLFM) has been adapted to characterize axial defects developing from damage in line pipe using J-integral-based [e.g. 17, 18] formulations since the 80s (e.g.,[19]). When simple models for axial cracking are adapted to possible crack growth from damage, care must be taken to address possible changes in driving force involving 1) the local stresses and strains including re-rounding effects, 2) wall thinning, 3) the flow response, 4) the effects of short crack mechanics considerations, and so on. Consideration also must be given to changes in the resistance to cracking (tearing) such as: 1) layered microstructural modification, 2) pre-strain effects, 3) differences in properties that relate to shallow cracks versus the deep-crack test geometries used to quantify resistance [10, 20]. Of these, the only major first-order uncertainty in the damage severity assessment framework developed in the PRCI PR3-9305 circa 2000 lay in quantifying the effects of rerounding, because while that work addressed this issue it did not do so adequately to quantify a generic solution [17]. For the present predictions, without the physically and geometrically nonlinear mechanics analysis done to address the concerns identified in PR3-9305, differences in controlling factors like diameter, D, and diameter to thickness, D/t, and properties oblige the use of experience-based trends developed for other cases, and empirical inference in reference to the axial displacements reported earlier. 5.2.1 Simplifications for Modern Tougher Steels: Mechanics and Modeling Implications

deep defects could survive to pressures the order of SMYS and above [4]. It also has shown that the failure response at such defects can be predicted [3], including whether failure will be collapse controlled versus fracture controlled, with the mode of failure discriminated as a function of pipe properties (i.e., geometry, yield stress, Y, ultimate stress, U, and fracture resistance expressed by CVN energy). Where collapse controls failure, quite high failure pressures like that observed in [3] can be anticipated. This occurs because even though the crack is initially sharp, the pipe toughness suffices to blunt the crack, such that the onset of stable tearing, growth, and instability leading to failure are deferred until quite high pressures. For the present damage growth predictions this means that initially sharp cracks formed in a transformed microstructure below the damage (as in Figure 4.1) will tend to blunt at the bottom in their transition into the untransformed structure below that layer, requiring quite high pressures for stable tearing. It also means that the eventual growth and instability require a larger crack driving force than found in damage databases dealing with less tough steels, where fracture controls failure, rather than collapse. Finally, the high pressures associated with failure under collapse conditions means that re-rounding occurs well prior to failure, which greatly simplifies the analyses. 5.2.2 Cracking Scenarios Leading to Immediate versus Delayed Failure

The above discussion and many cross-sections made through damage evaluated in field-failure analyses and for simulated damage indicate that 1) several types of cracking can develop, and that 2) cracking can evolve through stages in depth, both of which depend on the nature of the damage, the pipes properties, and factors that control the crack driving force such as pressure. Quite shallow (micro) cracks can form perpendicular to the principal stress particularly in hard brittle deposited material, or in a work-hardened microstructure that is severely worked (pre-strained) or transformed. Other circumstances favor the formation of shear cracks, which tend to grow deeper and remain stable, unless their growth carries them from the worked layer into the transition gradient and the base structure of the line pipe, which in the limit breaches the wall leading to immediate failure. Sliding contact can cause significant work hardening, and frictional heating, whereas normal contact is less so prone. Quite planar damage zones can cause failure, particularly where the damage implement leads to significant localized re-rounding. Evaluation of cracking evident in cross-sections through damage coupled with the nature of the strengthening mechanisms of steel and their influence on cracking in the work-hardened or transformed layer below damage suggests four cracking scenarios can be anticipated during or after rerounding [1]. Of these, crack growth to a critical depth at its current or a coalesced length during re-rounding in the wake of the damage underlies immediate failure (scenario 1), as can plastic collapse through the remaining ligament. Fracture

Prior PRCI work involving burst testing of pipe made of tough steels but absent mechanical damage showed that quite long

Copyright 2012 by ASME

mechanics and plastic collapse analysis adapted in the context of [1] provide the means to distinguish between this scenario and the three that underlie delayed failure or benign cracking, as detailed later in regard to the full-scale testing of the damage simulated as part of MD-4-1. Three additional scenarios can be anticipated leading to delayed failure or benign cracking. Benign cracking (scenario 4) cases occur where the damage develops no significant cracking in the hardened or transformed layer during rerounding to the maximum allowable operating pressure (MAOP). Significant, frequent pressure cycling is necessary to initiate fatigue cracking absent a local stress raiser for benign cracking. However, had that stress raiser been present during re-rounding, its presence is anticipated to otherwise nucleated cracking during re-rounding which thus indicates significant stress raisers are absent where benign cracking occurs. Corrosion and/or stress corrosion could likewise induce initiation, but such is true whether or not damage occurs the key difference being the coating is locally removed . The second scenario involves cracking into the low-toughness hardened layer, with the expectation that it was likely to continue through the full depth of that layer, but then promptly arrested in the tougher base pipe. Cracking introduced in this way can grow by fatigue, as well as by stress corrosion, or time-dependent stable tearing. However, because such cracking is still relatively shallow rather frequent or severe pressure cycling is indicated to be necessary to cause fatigue crack initiation absent a local stress raiser whose presence is likely for this scenario. The third scenario involves cracking that continues into the base pipe, but remains stable at the re-rounding pressure because the driving force for cracking is too small to overcome the steels inherent resistance. Cracking with this scenario also can grow by fatigue, by stress corrosion, or time-dependent stable tearing, with the clear potential for delayed failure in cases where the damage and initial rerounding occur at less than MAOP. Hydro-retesting also can activate such cracking. While as yet unproven by controlled testing under laboratory conditions, these scenarios are consistent with field observations particularly for benign cracking found adjacent to that causing a delayed failure, or in digs associated with features identified as concerns following ILI and related digs. Likewise, they are consistent with delayed failure due to time-dependent stable tearing wherein the damage and initial re-rounding are incurred at a pressure less than MAOP(e.g. [17]). 5.2.3 Depths of Subcritical (Stable) Cracking in Contact Zones

is significantly work hardened or transforms to un-tempered martensite, cracks nucleating in that layer are very likely to grow unstably through the layers full thickness. Inherently lower toughness steels are prone to further reduction of toughness due to the effects of pre-strain, as demonstrated in [1], established more broadly since (e.g.,[17]). However, such effects are relatively diminished in high toughness steels, such that modern X52 and X65 evaluated in the work of MD-4-1 is anticipated to show only marginal effects, if any. Pre-strain effects on pipe toughness were shown in the MD-4-1 project to be not significant . This coupled with the high toughness of the steels evaluated in MD-4-1 suggests cracking upon re-rounding would be contained within the brittle hardened layer, and blunt thereafter if the damaged pipes were pressurized to failure. Because the depths of the microcracks developed in the simulated damage (e.g., Figure 4.1) are the order of the scale of the blunting that forms at the base of the damage prior to incipient failure, the presence of such cracking during rerounding nominally adds nothing to the reported depths of the quite large gouges that formed in that simulated damage. And, because significant plastic strain develops en route to plastic collapse, residual stresses are secondary, which further simplifies prediction of failure at such damage to the analysis of failure at metal loss. It follows that the sizes of the gouges reported in terms of their lengths and depths as part of MD-4-1 can be directly used as the basis for failure prediction, without concern for microcracks with failure evaluated in this context based on metal loss failing by plastic collapse. Because the hardened layer has much reduced toughness as compared to the base pipe for almost any line-pipe steel, three potential delayed failure states form upon re-rounding [10]: 1. No significant cracking in the hardened or transformed layer (which precludes cracking into the base pipe steel), was asserted to be benign for historic gastransmission service (scenario 1) 2. Involving cracking into the low-toughness hardened layer, with the suggestion that it likely continued through its full depth and then quickly arrested in the tougher base pipe (scenario 2) 3. Cracking that continues into the base pipe that remained stable at the re-rounding pressure (scen. 3). As dictated by fracture mechanics, the cracking depth into the hardened layer or into the base pipe depends on the pipes toughness, and the severity of the crack-driving force [15]. Because the toughness in the hardened layer is low, and the response likely quite brittle if the steel transforms to untempered martensite, cracks nucleating in that layer are very likely to grow unstably through the layers full thickness. Once into the base pipe, because the toughness there depends on the vintage and the chemistry and processing involved in its production, the range of continued crack responses depends broadly on the pipe involved rather that the damage state. This can lead to significant variability in cracking behavior, and the failure pressure, which some might misconstrue as

As dictated by fracture mechanics, the cracking depth into the hardened layer or into the base pipe depends on the pipes toughness, and the severity of the crack-driving force (e.g., [15]). Following the logic of the prior subsection in light of Figure 4.1, because the toughness in the hardened layer is relatively lower, and the response likely quite brittle if the steel

10

Copyright 2012 by ASME

scatter, or poor quality data. Further on such cracking is noted in [15] and [17]. 5.2.4 Plastic-Collapse Predictions for Metal Loss due to a Gouge Because even initially sharp crack-like features blunt in higher toughness steels, plastic-collapse analysis for blunt (metal-loss) features is relevant for predicting failure pressure for the gouges created in MD-4-1. In such cases, the burst pressure, Pf, depends for a given defect depth on the mechanical properties of the pipe and the length of the defect, with the criterion PCORRC[21] used along with the above-noted empirical insight in the context of PR3-9305 to predict the burst test outcomes for all testing reported earlier. The same model, with the same properties has been used for each of the pipes, which is necessary to illustrate the predictive trends, and is justified by the fact that all pipes of a kind come from the same heat.

Figure 5.2 summarizes the predictions of failure pressure corresponding to the final defect depths, with the predicted outcomes shown on the y-axis as a function of the actual failure pressure shown on the x-axis. It is apparent from this comparison that the predicted outcomes for all experiments correspond closely to the actual results, largely within a scatterband comparable to the variability in the measured values of the UTS. Such accurate outcomes serve to validate the predicted values of af that are used later in the FCG analyses.

Figure 5.2 Comparing predicted with actual failure pressure (approach of PR3-9305 [10]) As these are plastic collapse controlled failures and due to the sizes of the damage features created for the testing in MD-4-1, quite high failure pressures are predicted consistent with the observed results. While good predictions are evident, without more comprehensive parametric numerical analysis of rerounding and its impacts on both the continuum concepts used to predict whether cracking initiates, and complementary fracture mechanics to address the onset of stable tearing and its growth, such models will remain only research tools. 5.3 Fatigue failure predictions a first

Figure 5.1. Predicted failure boundaries for MD-4-1 tests on X52 Figure 5.1 developed based on the PR3-9305 approach presents the failure boundaries shown on coordinates of pressure and defect length for the X52 pipe testing done in MD-4-1. The contours shown in this figure represent defect depth, normalized relative to the wall thickness, and serve to quantify the final defect depth, af. Whether leak or rupture (LvsR) is predicted is dealt with by independent analyses, the scope of which is beyond the present paper. Comparable analyses based on the same model, with the same properties used for each of the X65 pipes, have also been done for the X65 testing, but not included here to save space. It is evident from Figure 5.1 that the three results predicted for the X52 are rather good, in spite of this approach dating to circa 2000. In general the results are within the scatter evident in the pipes mechanical properties.

approach
Consider now initial outcomes from the FCG analysis. Without geometrically and physically nonlinear FEA to help understand the possible roles of damage-induced residual deformations and stresses, predicting fatigue response that can involve exponents that range from 1/12 to 1/5 on initiation parameters, and from 3 to 4 on propagation parameters can be challenging. Further complications exist in regard to the transition in properties in

11

Copyright 2012 by ASME

the layers that form below the damage implement, and in predicting the potential population of initial crack sizes due to damage. In spite of these complications, the fact that plastic collapse controls failure can be coupled with insight to quantify the plausible cracking population, and whether depth suffices in lieu of lengths and depths based on the literature and experience, as the basis for a preliminary model of FCG. In regard to the adequacy of depth alone, experience indicates that once the defect is long compared to its depth, its depth tends to dominate predicted life with little dependence on length. This tendency traces to the role of length and depth in quantifying the crack-driving force[15]. Lessons learned via LEFM FCG sensitivity studies done for critical nuclear piping in the late 1970s[19] provides further insight. Such trending indicates that four parameters control, including: the local stress range sensed by the crack tip; the threshold for growth including local closure and related issues; and the initial and final crack sizes. It follows that each of these parameters must be embedded in the efforts of MD-4-4. While the adequacy of depth can be in reference to the literature, without FEA characterization of re-rounding and other aspects that affect the local crack driving force it is necessary to infer their significance by contrasting blind predictive outcomes with the results of the full-scale tests. Thus, the approach starts with the assumption that cracking develops below the damage, and that such cracking can be quantified using linear-elastic fracture mechanics (LEFM), with gaps and complexity due to aspects like re-rounding effects bridged as the demand for more complex modeling become clear from such comparisons. In this context, above the threshold for FCG early work through the middle 1970s showed that there was little effect of the steel type on the FCG rate [20], which is denoted da/dN (where da is the incremental FCG per cycle and dN is the increment in number of cycles). 5.3.1 Estimating Life After Damage In the simplest terms, the service life beyond incurring the damage is given in the form:

include constraint, and involve the growth of much shallower cracking that is prone to short-crack effects due to plasticity closure(e.g. [23]), and other concerns including the threshold for deep cracks that is often nonconservative in applications involving shallower cracks. The last parameter essential to FCG prediction below damage is the value of ai the initial depth which was established empirically by trending observations from cross-sections like that in Figure 4.1. Consistent with the approach that addresses complexities in the crack driving force and other parameters based on the predictive outcomes, Equation 1 has been evaluated initially based on the observation that once a crack grows beyond a stress gradient that originates it, the crack driving force for continued growth quickly transitions to a form that ignores that gradient [15], and absent closure concerns also is independent of the wake of its growth [15]. That is, once the crack has grown marginally into the pipe below the damaged material at the gouge, its response can be quantified by that of a crack in a damage-free pipe whose depth is quantified relative to the original wall thickness rather than the bottom of the gouge. This expedient had been adopted on a preliminary basis in lieu of modeling the early growth from the gouge, as experience indicates that level of refinement is not justified absent knowledge of the role of re-rounding whose effects likely dominate that local response. Accordingly, for the damage depths noted earlier that ranged from about 13 to 49% of the actual wall thickness, if the observed microcracking to depths the order of 0.1 mm is added to the damage depths, then the initial crack depths are determined to range from about 1.1 mm (0.043 inch) up through about 4 mm (0.158 inch). While this is not inconsistent with fracture mechanics concepts, unless microcrack initiation occurs at the base of the gouge during damage, then the bottom of the damage would remain a blunt feature. Fatigue crack initiation could occur as the basis to justify continued FCG or stable tearing. Likewise, microcracking could occur upon re-rounding. Indicators of rerounding-induced cracking include empirical data from fullscale testing presented here, as well as analytically [10]. Accordingly, there is a physical basis to initially assess the remaining life solely in terms of FCG.. While physically justified, as becomes evident shortly the remaining lives predicted on this basis are strongly dependent on ai. Thus, if models like that being developed are to be useful in prioritizing ILI indications of damage, then the populations of defect depths that form in the wake of the damage must be better characterized. 5.3.2 Crack Driving Force Based on the above-noted construct, Equation 1 has been evaluated based on the nominal pressure histories used in the fatigue testing as reported earlier. For that analysis, the LEFM

af

ai

f K da

(1)

where K is the range of K determined from the pipelines operating conditions, ai is the depth of the cracking due to damage, and af is the final depth that comes from results like that in Figure 5.1. For present purposes the final crack depth parameter af follows from outcomes like that shown in Figure 5.1, with the value of af equal to either the critical depth when rupture is predicted, or to the remaining wall thickness when a leak is predicated. The f(K) in Equation 1 is empirically determined, which in simple terms is expressed in the usual Paris power-law [22] format: da/dN = Cr (K)n. Regardless of the form used, resistance is quantified using laboratory tests on simple test geometries, with n and Cr usually determined using a relatively deep crack growing in an opening mode under limited lateral constraint. In contrast, the FCG at damage can

12

Copyright 2012 by ASME

crack driving force K was functionally expressed as K = S Ck (f[a, 2c/a, a/t, 2c/W]) relative to an isolated semi-elliptical defect. The function denoted f[a, 2c/a, a/t] is quantified via handbooks (e.g. [24]) or by case-specific numerical analysis, while Ck is a constant whose dimensions are determined relative to the specific form of the function used, and also dependent on n. As evident in Equation 1, the value of K in FCG analysis is denoted K to indicate that it is the change in K due to the cycling that controls the outcome. The range of K (i.e. K) is defined relative to the change in the remote stress normal to the crack plane, S, which for the full-scale testing is proportional to the hoop stress and/or pressure. While at this initial stage in the modeling the value of K has been evaluated simply in regard to the cyclic pressure history, crack advance responds to the local stress range sensed at the crack tip. As axial cracks grow into the depth of the pipe wall, they experience the coupled effects of membrane tension and local bending due to bulging, such that the value of K very likely reflects some combination of these stresses. As such, a value of K based only on membrane tension (Km) could be viewed as a lower bound on K (equally an upper bound on life), while that based on membrane tension and local bending (Km+b) could be considered an upper bound on K. On that basis, both estimates of K were formulated in terms of the nominal pressure cycling, with these components of K based on work of Tada et al. [24]. In this context, the upper-bound form of K was comparable to that adopted by Hopkins circa 1982 [25], except in regard to his plastic zone correction, which for the present scenario affects a small reduction in life as compared to the relative effect of the initial depth. As alluded to earlier, possible modifications may be needed to account for aspects like rerounding and crack closure, and large-scale crack tip plasticity. As circumstances motivate the use of a more refined formulation for the components of K, these and other complexities will be incorporated. 5.3.3 The FCG Resistance Parameters As the cracking into the hardened layer evident in Figure 4.1 runs to the interface between that layer and the gradient into the base pipe, there is a chance that local residual stresses affect its behavior, as might differences in the hardness of the steel in that gradient. However, as noted earlier, growth above the threshold for FCG tends to be insensitive to microstructure, and residual stress levels measured remote to cracking will be altered by the cracking, and local crack-tip plasticity. Accordingly, the FCG resistance has been characterized simply in regard to the results such as that in [12], with n taken as 3.5 and Cr taken at 1.1x10-10 (in ksi in 0.5 units). While a typical value of the deep-crack threshold was taken at 4 ksi in 0.5, this parameter was not a factor for these analyses.

5.3.4 Results of the Fatigue Crack Growth Predictions for Cracking in Pipes 1 and 2 Table 5.1 summarizes the details of the fatigue testing relative to the lengths and depths of the defects, the pressure histories, and the observed lives to failure, while the table and Figure 5.3 present the corresponding life predictions. It is apparent from table 5.1 that one of the defects (gouges) was relatively shallow, while a second was relatively deep, with the remaining three more or less comparable. Two defects were quite long, with the remainder having comparable length. From a fracture mechanics perspective depth has a greater relative effect on driving force as compared to length such that the deepest defects would be anticipated to show the shortest lives, all else being equal. All defects experienced the same nominal pressure range (table 5.1), with four of the five tests done at high maximum stress relative to the usual MAOP for transmission service. In contrast, the test with the shortest life involves the lowest maximum stress explanation for this short life being the comment in table 5.1 that this defect experienced cyclic re-rounding. It is emphasized in this context that, while scatter does occur in fatigue testing, and the two tests with long defects did not show microcracks with MPI, this shorter life was anticipated because the cyclic pressure and peak pressure were selected a priori to maximize the effects of re-rounding.

Figure 5.3.a. Crack growth predictions for the defects in Pipe 1. Figure 5.3a presents results of predicted crack-growth response for Pipe 1, wherein six predictive trends are shown, while Figure 5.3b presents the predictions for Pipe 2, where four predictive trends are shown. Two trends are presented for each defect in each of the two pipes. That shown with data points presents results for Km, while that shown as the corresponding dashed line presents results for Km+b. As indicated earlier, if this simple approach to formulate K is viable, then the actual test results are anticipated to fall between the respective bounds for each of the five reported tests. The y-axis in this figure

13

Copyright 2012 by ASME

indicates crack depth below the pipes surface, as noted above (not the depth below the gouge), while the x-axis is the number of applied cycles. The scales adopted for this figure have been chosen such that the results for Pipe 2 can be shown on the same scales to facilitate comparison. For Pipe 1, predictions are shown for each of the three initial crack depths expressed relative to the wall thickness, which based on the above discussion cover values of a/t of 0.127 (1.1.3), 0.494 (1.2.3), and 0.264 (1.3.3). Note that the parenthetical sequences (1.1.3, etc) follow the same pattern reported earlier in this paper. For Pipe 2, predictions are shown for both of the two initial depths considered, which also based on the above discussion cover values of a/t of 0.246 (2.1.3), and 0.302 (2.2.3). All predictions have been terminated at the same relative depth set at 70% of the wall thickness, because as can be seen from Figure 5.3 there is nominally no change in predicted life beyond that point. Other depths of this same
Pipe # 1 1 1 2 2 Defect Type 1 2 3 1 2 Total Depth %t mm 12.6 49.3 26.3 21.5 26.4 0.99 3.89 2.08 1.94 2.38 Lengt h mm 136 105 321 165 319 Max Stress %YS 73 73 80 80 37

order also could be arbitrarily chosen, so long as the predicted outcome is unaffected. Figure 5.3.b. Crack growth predictions for the defects in Pipe 2. It is apparent that all predicted trends show the usual response, with the depth for the shallower cracking growing gradually at first and then accelerating as the cracking begins to sense the inside wall of the pipe (due to the so-called back-surface effect on K). Because the same values of Cr and n apply to all predictions, and a comparable nominal pressure range applies for all cracking, the outcomes for a given pipe reflect the same generic response with initial depth the only discriminator on the outcomes. This is illustrated in regard to Figure 5.3a by use of the horizontal line segments drawn between the adjacent

Nominal Stress Range %YS 35 35 36 35 25 bar 40 40 40 40 40

Life in Cycles Predicted Km Km+b 82379 29836 519 42 21251 4750 25889 5787 13127 2370 Actual 10869 5200 20494 17700 2007

Comments

Several unplanned interruptions Above range for re-rounding At maximum = 80% YS In the range for re-rounding

Table 5.1. Details and outcomes of the fatigue testing trends. The influence of differing values of K is also evident in Figure 5.3, with this influence almost as strong as that for the range of initial crack depths. The role of local bending in this context is very significant, causing a rapid acceleration in growth and much shorter predicted lives. Table 5.2 compares the predicted lives with the actual failure response of the damaged pipe, with results shown in the same format as Table 5.1 so that the influence of factors like the stressing and the defect sizes can be evaluated. Note that the tabulated predictions reflect outcomes for both estimates of K, which as noted earlier were anticipated as bounds on the lives developed in the full scale testing. The column denoted Km presents predictions based on the nominal range of the membrane stress, while that denoted Km+b presents the predictions that include both membrane and bending components. Values tabulated that best match the actual results are identified by an italic font. Recognizing that full-scale testing under otherwise identical conditions can lead to differences in actual life of a factor of two, but often larger, the observed and predicted lives might be viewed as comparable particularly given the simple formulations used to estimate K, and the uncertain initial crack depths embedded (for the three defects with microcracks evidence) in these predictions. Allowing for scatter, the expectation that these estimates would bound the actual data holds, with the exception of the deepest initial depth which underestimates reality by an order of magnitude or more depending on the prediction considered. Two of the predictions are best represented by the result that includes the bending component, while the other three are best represented by the membrane only prediction. While there are aspects that suggest these are reasonable predictions, which for some cases can be true, there are trends in these outcomes that point to the need to better understand when the effects of local bending become a controlling factor. Overlaid on that is the observation that the local effects of plasticity have been ignored in this predictive approach, as have the effects of re-rounding which clearly is observed in the outcomes for the testing of defect 2.2.3, as identified in the comments for that result in Table 5.1. 5.3.5 Sensitivity Analysis and Discussion Recognizing that reality for the full scale testing lies between the limits of Km and Km+b, a sensitivity study was done using an empirical blend of these, which for defect 1.1.3 led to a predicted life of 34,888 cycles, as the basis to illustrate the sensitivity of predicted life to the initial defect depth (Figure 5.4 on the same coordinate system as Figure 5.3, with scales to suit these outcomes). For this reference case ai was taken as 0.99 mm (0.039 inch). The results shown in Figure 5.4 clearly show the significance of the initial crack depth, as just more than a 51% increase in

14

Copyright 2012 by ASME

depth causes roughly a 13-fold decrease in fatigue life. Thus, it is clear that this parameter must be adequately characterized if useful guidelines are to be result from projects like the present. It is also apparent that if the initial size is accounted for the final size has much less influence, at least for these results and those in Figure 5.3, although this is not always the case. While the crack driving force involves much more than initial crack size, the strong dependence on initial depth coupled with the range of initial depths covered in Figures 5.3a and 5.3b suggests that a significant range of fatigue lives is anticipated from such predictions, as was observed.

be quantified and its evolution with cycling understood, which is likely best done empirically. Analysis of the region of and also below the damage in this testing indicates the presence of notches or grooves as origins for cracks, whose role should be assessed, along with the implications of the properties of the steels in such areas, and the local plastic zone. Experience suggests this is best dealt with analytically, supported by selected empirical trending relative to the geometry of the damage and the magnitude of the local stresses and strains [1]. Experience also that the growth of cracks in inelastic gradient (notch) fields [26] cannot be simply predicted by LEFM or NLFM unless the unique local fields and crack closure are addressed, but experience also suggests this may be a secondary factor to issues like re-rounding. Re-rounding effects are unique in that they can in some cases accelerate cracking, whereas other circumstances can slow it. As such, rerounding alone could account for the shifts between Km and Km+b, which make it a first-order concern. Systematic but limited analyses have considered various model options relative to the desirable use of nominal pressure cycling, including notch effects, plastic-zone effects [including a strip-yield form]) in an (EPFM J-based) approach (as outlined previously in [12]). Similar limited evaluation was made based on the empirical trends in the full-scale testing, with a view to quantify the stress cycling as it occurs local to the damage. Such results showed that predicted lives are reduced fractionally through use of EPFM forms and provision for the local plastic zone within a LEFM framework. However, quite significant shifts in predicted life are evident when a local estimate of the stress cycling in the vicinity of the damage was employed. Of these factors, order of magnitude reductions were possible in the context of a local NLFM driving force and when the shallow (so called short) crack effect [23] was addressed in an LEFM framework. It follows in this context that understanding of the local response is critical, as is the role of shallow versus deep crack fracture mechanics. 6 CONCLUSION As a summary, the present paper developed two aspects related to improving our fundamental knowledge about the failure mechanisms of dent and gouge defects. Experimentally, several outcomes concerning modern, tough steels were established on a database of 15 tests: Highly dynamic impact leads to the creation of a very hard superficial layer containing tooth material, and very prone to micro-cracking; its limited thickness 100 to 200 m deep confines these micro-cracks above a strain hardened layer; more investigations will be performed also on the deep dents with a less quick aggression For modern clean and very tough steels, moderate dent and gouge damage does not affect significantly the pipes burst strength; Severe

Figure 5.4. Illustrating the sensitivity to initial defect depth (relative to Defect 1.1.3). 5.4 Discussion As indicated above, the suitability of the relatively simple model that underlies the predictive outcomes in Figures 5.3 and 5.4 would be judged acceptable if its results were consistent with the full-scale testing. While the range of predicted lives was consistent with expectations in terms of initial defect depth, it is clear that the sequence of the shortest to longest life is not consistent with the mix of damage depths and effective local pressure cycles. On this basis it must be concluded that the simple model fails to capture all of the key aspects of fatigue crack growth from the damage for the pressure histories imposed, such as re-rounding. While there is a tendency to slightly over-predict life in some cases, which could be offset easily by slightly larger initial crack depths, there are also outcomes that underestimate life. Consequently, fundamental gaps exist in this predictive framework. There are several potentially competing factors complicating understanding the drivers for these trends, whose relative significance can only be identified by systematic analysis of each factor. Consideration of the literature indicates several plausible factors can lead to accelerated FCG rates, while others can lead to shorter lives. First, the microcrack depth population must

15

Copyright 2012 by ASME

gouging was found to be the only case of significant drop in burst pressure below the flawless pipe For deeper dents, ~5% residual depth, for which the re-rounding process involves larger displacements above the internal pressure during denting, there is a strong influence of the mean level of the pressure cycling range: if this is immediately above the pressure during denting, a large cyclic re-rounding will lead to a short fatigue life; conversely, if the pressure range is well above the dent pop-up pressure range, then the fatigue life will be longer, and displacements will be smaller. This finding has not been verified on shallower dents yet, as it used a novel measurement technique. A first fatigue modeling approach showed that, based on limited data, it is possible to characterize the fatigue response of line pipe damaged by dents and gouges, but it is clear that this requires a viable assessment of the initial depth of the cracking, and can in some pressure-contact circumstances require an understanding of the local effects of re-rounding in the wake of the damage implement.

[4]

[5]

[6]

[7]

[8]

[9]

Experimental work will go on with finishing metallurgical characterizations and undertaking a similar work on two different vintage pipes, that will add 12 more defects to the data-base: 4 cases x 3 defects / case. Modeling work must focus on better quantifying the local rerounding and its influence on the stress-cycle experienced at the crack. Once this is complete, if needed the work will more broadly consider other local nonlinearities. 7. ACKNOWLEDGMENTS The authors hereby acknowledge the funding received from PRCI and DOT to pursue this work on dent and gouge mechanical strength, and the international coordination with EPRG. All individuals that contributed valuable work via the PRCI project teams, or the DOT project team, and the two contractors teams should also receive our recognition. 8. REFERENCES [1] Anon. 1970 2007 Gas Pipeline Incidents 7th Report of the European Gas Pipeline Incident Group. www.egig.eu (December 2008). Anon. All reported gas transmission pipeline incidents. Interpretation by the authors (2011). http://primis.phmsa.dot.gov/comm/reports/safety/. B. N. Leis, R.B. Francini Line Pipe Resistance to Outside Force. Volume 2, Assessing Serviceability of Mechanical Damage, PR-3-9305, Catalog L51832, PRCI, Battelle Memorial Institute (1999)

[10]

[11] [12]

[13]

[14]

[15]

[2]

[16]

[3]

[17]

Chang-Kyung Oh, Yun-Jae Kim, Jong-Hyun Baek, Young-Pyo Kim and Woosik Kim. A phenomenological model of ductile fracture for API X65 steel. International Journal of Mechanical Sciences, 49, p1399-1412 (2007). Noaki Fukuda, Naoto Hagiwara, Tomoki Masuda. Effect of prestrain on tensile and fracture toughness properties of line pipes. 4th International Pipeline Conference, September 29-October 3 (2002). Hai Qiu, Manabu Enoki, Kazuo Hiraoka and Teruo, Kishi. Effect of prestrain on fracture toughness of ductile structural steels under static and dynamic loading. Engineering Fracture Mechanics, 72, pp 16241633 (2005). Marshall, P. W., Strategy for Monitoring, Inspection and Repair for Fixed Offshore Platforms. in Structural Integrity Technology, ASME. pp. 97-110 (1979). Yukawa, S., ASME Nuclear Code Applications of Structural Integrity and Flaw Evaluation Methodology. in Structural Integrity Technology, ASME, pp. 29-38, (1979). Leis, B. N., Mayfield, M. E., Forte, T. P., and Eiber, R. J. Influence of Initial Defect Distribution on the Life of the Cold Leg Piping System. in Mechanics of Nondestructive Testing, Plenum Press, , pp. 325-342, (1980). Bolton, B, Semiga V., Tiku, S. Dinovitzer, A. And Zhou, J. Full Scale Cyclic Fatigue Testing of Dented Pipelines and Development of a Validated Dented Pipe Finite Element Model. Proceedings of the 8th international Pipeline Conference, IPC2010-31579, (2010) Anon., ANSYS 11.0 Reference Manual, ANSYS .Inc. (2009) Leis, B. N. and Francini, R. B. Line Pipe Resistance to Outside Force Volume II: Assessing Serviceability of Mechanical Damage. PRCI Report PR3-9305, (2000). Kiefner, J. F., Maxey, W. A., Eiber, R. J., and Duffy, A. R. Failure Stress Levels of Flaws in Pressurized Cylinders. Progress in Flaw Growth and Fracture Testing, ASTM STP 536. pp. 461-481 (1973) Leis, B. N., Brust, F. W., and Scott, P. M. Development and Validation of a Ductile Flaw Growth Analysis for Gas Transmission Line Pipe. PRCI Catalog No. L51543, Pipeline Research Council International (June 1991). Marshall, P. W. Strategy for Monitoring, Inspection and Repair for Fixed Offshore Platforms. in Structural Integrity Technology, ASME, pp. 97-110 (1979). Yukawa, S., ASME Nuclear Code Applications of Structural Integrity and Flaw Evaluation Methodology. in Structural Integrity Technology, ASME. pp. 29-38 (1979). Broek, D. (1974) Engineering Fracture Mechanics, Nordhoff.

16

Copyright 2012 by ASME

[18]

[19]

[20]

[21]

[22] [23]

[24]

[25]

[26]

[27]

[28]

Hahn, G. T., Sarrate, M. and Rosenfield, A.R. (1969) Criteria for Crack Extension in Cylindrical Pressure Vessels. I J Frac Mech, Vol.5 pp. 187-210. Leis, B. N. and Hopkins, P. (2004) Mechanical Damage Gaps Analysis. in Volume I -Pipeline Technology, Science Surveys Ltd. pp 1921 1943; but most complete as (2002) PRCI Report PR3-0277. PRCI Cat L52013. Leis, B. N., (2001) Hydrostatic Testing Of Transmission Pipelines: When It Is Beneficial and Alternatives When It Is Not. PRCI Report PR-3-9523, Pipeline Research Council International: see also Olson, R. J., Narendran, V. K., Leis, B. N., Kilinski, T. J., Scott, P. M., and Gertler, R. C. Full-Scale Testing to Validate PAFFC and Develop Data to Assist Evaluating the Benefits of Hydrotesting, Appendix 7 of this report. Leis, B. N., Mayfield, M. E., Forte, T. P., and Eiber, R. J. (1980) Influence of Initial Defect Distribution on the Life of the Cold Leg Piping System, in Mechanics of Nondestructive Testing, Plenum Press. pp. 325-342. Rolfe, S. T. and Barsom. (1977) J. M., Fracture and Fatigue Control in Structures, Prentice-Hall. Stephens, D.R., Leis, B.N., Kurre, J.D., and Rudland, D.L., January (1999) Development of an Alternative Failure Criterion for Residual Strength of Corrosion Defects in Moderate- to High-Toughness Pipe. PRCI Cat. No. L51794,. Paris, P.C., Gomez, M. P., and Anderson, W. E., A Rational Analytic Theory of Fatigue, The Trend In Engineering, Vol. 13, pp. 9-14, January 1961. Leis, B. N., Kanninen, M. F., Hopper, A. T., Ahmad, J., and Broek, D., (1986) Critical Review of the Fatigue Growth of Short Cracks. Engineering Fracture Mechanics, Vol. 23, No. 5. pp. 883-896. Tada, H., Paris, P. C., and Irwin, G. R., Stress Analysis of Cracks Handbook, Pre-publication release, 1972 (third edition now available, ASME Press, 2000) Hopkins, P. and Carins, A., (1981) A Fracture Model to Predict the Failure of Defects in Dented Line Pipe. British Gas, EPRS R2382. Leis, B. N. (1985) Displacement Controlled Fatigue Crack Growth in Inelastic Notch Fields: Implications for Short Cracks. Engineering Fracture Mechanics, Vol. 22, No. 2 pp. 279-293: see also Gowda, C.V.B., Topper, T. H., and Leis, B. N. (1972) Crack Initiation and Propagation in Notched Plates Subject to Cyclic Inelastic Strains", International Conference on Mechanical Behavior of Materials, Kyoto, Japan, Vol. II. pp. 187-198.

17

Copyright 2012 by ASME