Sie sind auf Seite 1von 12

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

IPC2012-90504

Root Cause Analysis of Dent with Crack: A Case Study


Udayasankar Arumugam, Ming Gao, Ravi Krishnamurthy Blade Energy Partners Houston, Texas, USA Rick Wang, Richard Kania TransCanada Pipeline Limited Calgary, Alberta, Canada

ABSTRACT A combined caliper and tri-axial MFL in-line inspection (ILI) reported a bottom side 2.7%OD dent associated with 76% metal loss. The reported dent depth is well below the 6% limit while strain analysis of this ILI dent profile showed a maximum equivalent strain of 17.4%, which exceeds the 6% strain limit for gas pipeline. Due to the high dent strain level, the raw signals of metal loss were revisited, which indicated this associated metal loss appears to be a crack rather than corrosion. In-field investigation revealed that this dent is indeed associated with branched cracks both at internal and external pipe surface but no leak was detected. The primary objective of this case study is to determine the cause for cracking in the dent. As part of this study, detailed investigation was performed including LaserScan based strain analysis, lab pressure-cycle testing and fracture surface examination. An attempt was made to quantify the plastic strain damage of this dent and its susceptibility to cracking using the existing plastic damage models, namely, ductile failure damage indicator (DFDI), strain limit damage (SLD) and minimum elongation limit criterion. The investigation showed that the internal cracks were formed at the time of indentation while the external cracks formed by spring-back (elastic rebounding) due to the removal of rock constraint. Full size fatigue testing of this cracked dent showed leak failure modes rather than rupture. In this paper, the approach, results and the findings are summarized and discussed.

NOMENCLATURE ABAQUS - Finite element analysis software DFDI - Ductile Failure Damage Indicator Eqv - Equivalent FEA - Finite Element Analysis ID - Inside Diameter ILI - In-line Inspection LVDT transducer Linear Variable Differential Transformer, a sensor used to measure linear displacement MAOP - Maximum Allowable Operating Pressure MFL - Magnetic Flux Leakage OD - Outside Diameter PEEQ - Equivalent plastic strain SLD - Strain Limit Damage factor SF - Safety factor SEM - Scanning Electron Microscopy S4 element - 4 node shell element D , k - Damage occurring during the kth load increment Dform - Damage occurring during the forming o - Critical true strain of the material eq - Equivalent strain f - Failure strain

Copyright 2012 by ASME

h - Strain in the hoop direction l - Strain in the longitudinal direction L,k - Max permitted local total eqv plastic strain at the kth load increment Lu - maximum of m2, m3 and m4 m2, m3, m4 and m5 - Values calculated using material properties eq - Von Mises stress - Principal stress in the 1 direction at a point of interest for the kth load increment - Principal stress in the 2 direction at a point of interest for the kth load increment - Principal stress in the 3 direction at a point of interest for the kth load increment - Von Mises equivalent stress at a point of interest for the kth load increment peq,k - Change in total eqv. plastic strain drain during the kth load increment. m - Average of three principal stresses P = Pmax Pmin R ratio = Pmin / Pmax INTRODUCTION TransCanada Pipeline Limited (TCPL) inspected[1] the 30-inch natural gas pipeline segment with a Combo tool (combined caliper and tri-axial MFL). The inspection reported a bottomside dent having a depth of 2.7%OD associated with 76% metal loss. The reported 2.7%OD is well below the 6% depth limit as per ASME B31.8[2], however, the metal loss is significant. The pipe and dent attributes are given in Table 1. To investigate the nature of metal loss, the raw signals of caliper and tri-axial MFL were reviewed, and found that the tri-axial MFL signals appeared to be associated with a sharp profile of deformation and, possibly, an indication of gouge or crack, rather than corrosion. To confirm this, further exercise was made to determine the strain level of the dent using the ILI dent profile data. The calculated maximum equivalent strain is 17.4%, which is well above the ASME B31.8 6% strain limit for gas pipeline. Combining the findings from MFL signal evaluation and strain analysis suggested that the metal loss feature most likely associated with cracks. In-field investigation indeed revealed branched cracks on the OD surface of the shallow dent. However, no leak was found during in-ditch examination.
e ,k

In the paper, the dent strain was first assessed using ILI and field LaserScan profile data, and compared with that obtained from finite element analysis (FEA). The plastic damage models are then applied to quantify the degree of plastic damage and susceptibility to cracking. The full scale fatigue cycling testing was performed based on the historical operating cycles. Finally, fractographic analysis is performed to determine cracking sequence and fracture mode. The results are presented and discussed. DENT STRAIN ASSESSMENT Recently, strain based assessment models[3-6] for dents are widely accepted and used in the pipeline industry to determine the dent severity and also for prioritization purposes. These strain models often utilize the dent axial and circumferential profiles reported by multi-channel geometry tool or in-ditch LaserScan measurement. Implementation of a strain based assessment, either using the non-mandatory ASME B31.8 Appendix R, or other improved methods, requires accurate dent profile data. The calculated strains from the above equations are true strains. In this section, three different dent profile measurements are considered, namely, ILI data, manual grid data and LaserScan data. Strain analysis was conducted using the dent profile data obtained from the above three measurement techniques and compared against each other. ILI Dent Profile Data: Table 1 lists the dent attributes reported by the combo tool. Table 1: Dent associated with metal loss and its attributes
Pipe Attributes Feature Identifier Pipe Wall Material MAOP OD Thickness Grade (mm) (mm) (kPa) 762 9.5 6895 359 MPa (X52) Dent Attributes Construction Depth of Clock Depth Length Width Year Cover Position (cm) (%) (mm) (mm) h:mm 1958 90 2.7 120 97 6:15

1, k

2 ,k

3 ,k

DNT 1783

Figures 1 and 2 show the screen shot of this dent feature with its axial and circumferential deformation profiles, respectively.

Figure 1: Screen shot showing dent axial profile

Copyright 2012 by ASME

Figure 4: MFL signal screen shot showing sharp and strong characteristics of magnetic flux leakage Figure 2: Screen shot showing dent transverse profile These profile plots indicate the dent seems to be narrow and sharp at the dent apex, implying that the dent may have high strain at its apex location. The strain distribution and severity of the dent can be determined using the reported profile data. It is noted that the resolution of ILI tool with respect to the dent size and geometry plays a vital role in determining the dent severity. An accurate measurement of dent profile leads to an accurate prediction of strain value. The resolution of this ILI dent data is 6mm x 35mm (1:6 ratio) in the axial and transverse directions, respectively. The strain analysis was performed on this data utilizing a point-to-point strain analysis tool[7]. The calculated maximum equivalent strain (true strain) for this dent is 17.4%, which is at the dent apex location. This true strain level is beyond the engineering strain at UTS of the material. The calculated equivalent strain is much higher than the 6% strain limit for plain dent in gas pipelines. Figure 3 shows the maximum eqv strain plot using a modified equation. In-field Manual Grid Measurement: Following the recommendation, TCPL conducted an in-field investigation. Before excavation, the line pressure was reduced to 20% of MAOP. The pipe was found to be in contact with a rock at this dent location and the rock was removed. The rock appears to be sharp at the contact point with the pipe. The in-field examination indeed revealed three cracks originated near the apex of the dent, Figure 5.

Figure 5: Picture showing dent associated with crack The in-field UT inspection [8] reported that the associated crack is ~40mm long with crack depth of 8mm, i.e., about 84% nominal wall thickness. There was no leak detected during the excavation. In addition, the dent profile was manually mapped with a 10mm x 10mm grid, Figure 6. The dent depth was measured using the pit gauge and also with profile gauge plot was taken at deepest dent location. It should be noted, the in-field dent profile was measured from OD while the ILI profile was measured from ID.

Figure 3: Maximum equivalent strain (modified eq) plot Since this bottom side dent has high strain, it is possible that the dent could contain a crack or gouge. A review of the MFL raw signal showed sharp and strong signal characteristics of magnetic flux leakage, Figure 4. The predicted 76% metal loss is unlikely associated with corrosion, rather cracks or gouge. Therefore, the dent was recommended for further investigation.

Copyright 2012 by ASME

Figure 6: Picture showing the manual grid measurement Later, the grid readings are converted into the electronic format and used in dent strain tool to estimate the dent strain. The maximum dent depth reported in the manual grid method is 11.5 mm, which is approximately 50% less than that of ILI reported depth (21mm). The difference in the dent depth could be mainly attributed to the re-rounding effect due to rock removal. The calculated maximum eqv strain is 24.5% using the manual grid data, which is higher than ILI strain value. This difference is due to the narrow dent apex caused by spring-back effect when constraint was removed during excavation. At this point of time, it is unknown whether the crack existed long ago or formed on OD surface when rock was removed. Moreover, the in-field inspection did not report any internal crack. Since the dent is associated with cracks, the pipe joint was cut-out and removed for further investigation and failure analysis. LaserScan 3D Mapping: Both ID and OD surfaces of the pipe cut-out were sand-blasted as per the NACE No.2 specification [9] for LaserScan mapping. A 3D laser mapping of this dent with crack was undertaken to accurately map both ID and OD of the dent geometry using a portable 3D LaserScan system. Figure 7 shows the 3D laser scan image of this dent.

Figure 8: Dent axial profile plot at OD surface

Figure 9: Dent axial profile plot at ID surface The strain analysis was performed on these two LaserScan profiles. The calculated maximum eqv strain values for OD and ID dent profiles are 22.2% and 16.0%, shown in Figures 10 and 11, respectively.

Figure 7: 3D LaserScan of dent with crack Figures 8-9 show the axial profile extracted from LaserScan at OD and ID surfaces, respectively, with axial data resolution of 5mm and circumferential resolution of 2deg (13mm).

Figure 10: Maximum eqv strain plot (OD dent profile)

Copyright 2012 by ASME

Figure 11: Maximum eqv strain plot (ID dent profile) Comparison of Dent Measurement and Strain Result: The dent attributes such as depth, length, ad profiles are measured with three different techniques, namely, ILI, in-ditch grid measurement and the LaserScan. The ILI reported dent depth of 2.7%OD and its length and width of 120mm and 97mm, respectively. This ILI measurement is based on ID surface with internal pressure and constrained position. However, the in-situ measurement is based on OD surface after rock removal but at reduced internal pressure. LaserScan measurement was made at the laboratory without internal pressure and the dent was unconstrained. Table 2 presents the comparison of dent dimensions.

Figure 12: Axial ID profiles ILI vs. LaserScan

Table 2: Comparison of dent dimensions by different measurement techniques

Figure 13: Circumferential ID profiles ILI vs. LaserScan Table 3 lists the maximum equivalent strain values using the dent profiles obtained from three different measurements. The reported dent length and width have some differences by various measurement techniques. The dent depth predicted during the excavation (after rock removal and under reduced pressure of 20% MAOP) was 1.3%OD (11.5mm), which is about 50% lower than that of ILI prediction. The decrease in dent depth could be due to the spring-back effect of the dent by releasing the elastic energy when the rock was removed. Figures 12 and 13 show the comparison of the deepest axial and circumferential profiles between the ILI and internal LaserScan. Table 3: Comparison of strain estimated using profiles measured by different techniques

Copyright 2012 by ASME

The maximum eqv strain of this dent using different measurement techniques consistently showed greater than 6% strain limit. The maximum eqv strain obtained from in-ditch grid measurement is almost the same as LaserScan OD strain value, indicating a very good match. The dent spring back effect and pressure difference at the time of various measurements caused the dent shape change leading to the difference in strain values, which are discussed below. The maximum eqv strain based on ILI dent data is 17.4%, which is higher than that of ID LaserScan data (16%). This difference could be due to a slight alteration of dent shape at the dent apex due to spring-back effect. It is also possible the ILI profile data may have under-estimated strain due to lower resolution than that of LaserScan in the circumferential orientation (35mm ILI vs. 13 mm Laser). The maximum eqv strain for OD LaserScan dent is 22.2%, which is higher than that of ID LaserScan (16%) and ILI data (17.4%). This is due to the change of OD axial profile shape (see Figure 14), which is sharper at the apex location than that of ID LaserScan and ILI axial shapes. As a result, higher the axial bending strain and hence, higher the estimated maximum eqv strain for OD LaserScan.

Minimum Elongation Limit criterion. A brief discussion on the three failure criteria and their application are given in this section. An attempt was made to calculate the DFDI and SLD using finite element analysis and compared against failure criteria to determine the susceptibility of the cracked dent to crack initiation. Material testing was performed to determine the critical strain limit of this pipe material. Again, it should be noted that all strains used here are true strains, not engineering strains. Ductile Failure Damage Indicator (DFDI): In the metal forming and solid expansion of tubular industries, DFDI model is adopted as ductile failure criterion, which is based on tri-axial stress field, equivalent Von Mises stress and critical strain of the material. This failure criterion is derived from the concept that ductile failure results from initiation, growth and coalescence of voids on a micro scale, and formation of cracks during large plastic deformation. Hancock and Mackenzie[10] in the mid-70s followed Rice et als[11] work, and proposed a reference failure strain, f, i.e., a strain limit for ductile failure.

f = 1.65 o exp(

3 m ) 2 eq

(Eq.1)

where, m = average stress of three principle stresses in a triaxial stress field, and eq = equivalent Von Mises stress. The ratio of m/eq represents the tri-axiality of the stress field, and o = critical strain of the material, a material property measured by uniaxial tension testing, which is 51.2% for this pipe material. The total plastic damage (Ductile Failure Damage Indicator), Di, is the integral of the following equation:
eq

Di =

d eq

(Eq. 2)

Di value ranges from 0 (undamaged) to 1 (cracking). By the definition, ductile failure or failure of dent (cracking) will occur when Di 1. Strain Limit Damage (SLD): ASME Boiler & Pressure Vessel Code[12] Section VIII, Division 3 recommends an equation that estimates the accumulated damage as strain limit damage (SLD) using elastic-plastic finite element analysis results. Additionally, Section 8 has developed approximations for material properties based on specified minimum reduction in area and elongation to failure. These material properties are incorporated into equations below. The total strain limit damage, Dt is the summation of accumulated damage and is given by:

Figure 14: ID and OD axial profiles (LaserScan vs. ILI)

PLASTIC DAMAGE MODELS FOR DENT SEVERITY EVALUATION Dents in pipeline are permanent deformations caused by plastic damage. Therefore, it is appropriate to apply the plastic damage models to determine its severity and susceptibility to crack initiation. Three plastic damage models are available in the industry to deal with the possibility of crack initiation due to local plasticity, namely, Ductile Failure Damage Indicator (DFDI), Strain Limit Damage (SLD), and

Dt = Dform + D ,k 1.0
k =1

(Eq. 3)

Copyright 2012 by ASME

where,

D ,k =

peq ,k

L ,k

(Eq. 4)

L ,k

m5 1,k + 2 ,k + 3,k 1 3 1+m 3 e ,k = Lu e 2

(Eq. 5)

In order to calculate the DFDI and SLD, finite element analysis should be conducted to extract three principal stresses and the equivalent plastic strain at every node on the dent deformation region. However, a simplified approach has been developed to estimate the upper and lower bound of DFDI directly for dent profile without FEA, and will be discussed later. Finite Element Analysis: A 3D pipe finite element model with 30 OD is created using the general purpose finite element software, ABAQUS[13]. The pipe model is meshed with 4-node shell element (S4 element). Appropriate boundary conditions are applied on the model to avoid rigid body motion. Later, the pipe wall thickness and the elastic-plastic material property are assigned to the pipe model. The true stress-true plastic strain curve was obtained from the material testing. Since the ILI data had low transverse resolution of 35mm, ILI data was not used for FEA. The FE node and element size (length) are maintained same as LaserScan ID resolution of 5mmX5mm. This helped one-to-one mapping of dent displacement data as a loading condition. An elastic-plastic finite element analysis was performed with the objectives of determining the maximum equivalent plastic strain and principal stresses under static condition. Since ILI data produced spurious strain spike in FEA (due to low transverse data resolution), no attempt was made to calculate the DFDI and SLD. However, LaserScan internal profile produced a better FEA stress/strain distribution around the dent region. It should be noted that the LaserScan was performed at laboratory conditions with a reported dent depth (~1.4%OD), which is half of the ILI reported depth (2.7%OD) due to springback effect. In order to estimate the plastic damage of this dent, the high resolution LaserScan internal profile was scaled by 1.7 factor to match closely with the low resolution ILI data, as shown Figure 15. This scaled-up LaserScan ID profile was used in FEA to estimate the maximum eqv plastic strain (PEEQ) and compared to geometric strain obtained from the point-to-point strain tool. In addition to this, the DFDI and SLD values were calculated at all nodes using the FEA result.

Figure 15: LaserScan internal profile Before and after scaling and compared with ILI profile Figure 16 shows the equivalent plastic strain (PEEQ) distribution at the inside surface of the dent.

Figure 16: Equivalent plastic strain distribution (PEEQ) Table 4 lists the FEA maximum equivalent strain and the geometric strain obtained using the LaserScan data. The difference between the geometric strain and FEA strain is less than 8%, implying the good estimation by the point-to-point dent tool using the profile data. The calculated maximum DFDI and SLD values using the FEA result are 1.1 and 1.52 respectively, which are greater than 1.0, indicating that this dent was susceptible to crack initiation at the internal surface (ID) of the dent.

Copyright 2012 by ASME

Table 4: Maximum eqv strain, DFDI and SLD

where eq and 0 are the maximum equivalent strain predicted using ILI dent profile (geometric strain) and the material critical strain, which is 51.2% for this pipe material obtained from the material testing. Similarly, simplified equations can be derived using the SLD equation (Eq.3). The simplified SLD equations for this X52 pipe material grade are given below.

SLDupperbound =
Figure 17 shows the maximum principal stress contour plot with the probable crack propagation direction and also compared with the actual crack path. The FEA predicted crack path had good agreement with the actual crack initiation, validating the appropriateness of the plastic damage models. In addition to this, the reaction force was also extracted, which is 27 kips, indicating a very high external force might have caused this dent deformation, possibly during construction.

(0.2248)
eq
0.4308

eq

(Eq. 8)

SLDlowerbound =

(Eq. 9)

Using the above simplified equations and geometric strain obtained from LaserScan internal profile, the upperbound and lowerbound of DFDI and SLD values are calculated. Table 5 lists the comparison of DFDI and SLD obtained from FEA method and simplified equations without FEA. The upperbound values of DFDI and SLD are in good agreement with FEA calculated values. This demonstrates that the DFDI and SLD can be used to describe the pipeline dents susceptibility to crack initiation. Also, the simplified upperbound equations demonstrate that DFDI and SLD values can be calculated using the dent strain (geometric strain) obtained from high resolution ILI or LaserScan profile and material critical strain without running the complex FEA calculation. Table 5: Comparison of DFDI and SLD obtained from simplified equation and FEA

Figure 17: Maximum Principal Stress Plot (indicated probable crack path) vs. actual crack path It should be noted that the FEA result is essential to calculate the DFDI and SLD factors. However, it is not practical to run complex FEA for all dents reported by ILI to evaluate their plastic strain severities and susceptibilities to crack initiation. Therefore, an attempt was made to simplify the DFDI and SLD equations (Eq.1 and Eq. 2) using two extreme conditions, namely, (1) Assuming the bi-axial loading (10, 2=1, 3=0), which gives the upperbound DFDI value, and (2) Assuming the uni-axial loading condition with 10, 2=0, 3=0, which gives the lowerbound DFDI value. For thin-wall pipe under internal pressure with the above two conditions, the DFDI equation (Eq.1) reduces to the following:

Minimum Elongation Limit Criterion: A study performed by Francini and Ghodsi[14] proposed an alternative strain limit (Eq. 10) to ASME 31.8 for plain dents, using the minimum specified elongation given in API 5L:

eq

o
SF

ef SF

(Eq. 10)

DFDI upperbound

eq = 0
1.65

(Eq. 6)

where o and f are the true fracture strain and the specified minimum elongation, respectively, and SF is a recommended safety factor, SF=2. The equivalent strain (eq) is calculated using equation 11, which is originally proposed by Lukasiewicz et al[3]:

DFDI lowerbound

= eq o

eq =
(Eq. 7)

2 2 h + h l + l2 3

(Eq. 11)

As per API 5L, the minimum elongation to failure for X52 grade material is 21%. Applying a safety factor of 2, the

Copyright 2012 by ASME

alternate strain limit is 11.5% for a plain dent. The calculated eqv strains for this dent with crack are much higher than 11.5%, indicating a dent strain level is not acceptable for inservice but it did not predict the condition for crack initiation. In brief, this dent consistently demonstrated a failure condition (crack initiation) by all three plastic damage models and its applicability to the pipeline dents. Simplified DFDI and SLD equations without FEA can also be used to prioritize the ILI reported dents in the order of dent cracking as severity. PRESSURE CYCLE TEST To determine the remaining fatigue life and fracture mode (leak vs rupture) of the dent, a full scale pressure cycling test was performed. Before conducting the pressure cycle test, the pipe cut-out containing the dent with crack was examined with MPI and UT shearwave. Table 6 presents the OD and ID breaking crack dimensions. Table 6: OD and ID breaking crack and its attributes
Crack Parameter Length (L) Width (W) Depth (d) OD breaking inches 1.50 0.75 0.07 mm 38.10 19.10 1.80 ID breaking inches 2.00 1.75 0.21 mm 50.80 44.50 5.30 Ligament distance inches 0.11 mm 2.80

A pressure cycle test was carried out on this dent at Texas A&M Laboratory. Before performing a laboratory pressure cycle test, the pipe and the dent regions were hardness tested using GE PII portable hardness test equipment to locate if any local hard spots. The hardness test result indicated no hard spot in the dent region. The actual operating pressure data was analyzed with Rainflow counting method and Miners rule. The calculated maximum pressure is Pmax= 6783 kpa (984 psi), Pmin = 4667 kpa (677 psi) and the equivalent number of cycles (Neqv) is 3.7 cycles per year. Based on the Rainflow result, two different pressure cycle spectrums were developed for the test program. Figure 19 shows a pressure spectrum with relatively low aggressive cycles, which was attempted first. Since no leak or rupture happened, high aggressive pressure cycle spectrum (as shown in Figure 20) was attempted later.

Also, a leak test using color dye was performed from dent ID surface and no leak was detected, indicating no through-wall crack. A minimum ligament distance of ~3mm was reported between the internal and external crack. Figure 18 shows a close-up view of (a) OD crack, and (b) ID crack.

Figure 19: Low aggressive pressure cycle spectrum #1 (P = 307 psi & R ratio = 0.7)

Figure 20: Aggressive pressure cycle spectrum #2 (P = 885 psi & R ratio = 0.1) Figure 18: Picture showing the OD and ID cracks

Copyright 2012 by ASME

The pipe specimen was instrumented with nine strain gauges and LVDT to monitor the strain level and the displacement at the dent apex. Figure 21 provides a view of the test setup prior to testing.

Figure 21: View of pipe with cracked dent prior to pressure testing As described above, the pipe cut out was tested first with pressure spectrum shown in Figure 19. No significant leak or rupture was observed even after 8 hours (92 cycles) under this pressure spectrum and the estimated fatigue life is more than 25 years. The test was temporarily stopped and prepared for aggressive pressure cycle (Figure 20). A little water pool by the leak was seen at the end of 8th cycle and when the pressure was increased at the beginning of the 9th cycle, a stream of wter ejected from the crack C as shown in Figure 22, indicating the cracked dent failed by leaking.

Figure 23: The coupon broken at liquid nitrogen temperature for macro and micro examination Fracture surfaces were examined before and after cleaning using a stereoscope at low magnifications. The ID surfaces of Cracks A, B and C showed consistently dark stains but did not show on OD surfaces, indicating that the ID cracks have existed for a long period of time while OD cracks appeared to be freshly rusted and formed more recently, possibly when rock was removed during excavation. Among the three cracks (A-C), Crack C leaked first during the laboratory pressure cycle test. Steroscopic and SEM examinations were performed first on Crack C. Three distinct fracture surface regions were observed: the ID crack (region 1), OD crack (region 2) and mid wall or crack propagation (region 3), as shown in Figure 24. These regions were examined in detail with SEM.

Figure 22: Stream of water ejected from crack C Pressure cycle spectrum #2

FRACTURE SURFACE EXAMINATION After the pressure test, 3ft long pipe was cut out for material testing and fracture surface examination using SEM. A small coupon that contains the dent with cracks was extracted and notched along the crack lines up to near the crack tips. The coupon was broken at liquid nitrogen temperature to expose the fracture surface. Figure 23 showed the coupon after the break with labels of cracks, ID and OD surfaces.

Figure 24: Crack C fracture surface showing 3 regions The SEM examination of region 1 (ID surface) showed features of dimples and lot of corrosion pits; part of the pits may be due to cleaning (see Figure 25). The dark stain and

10

Copyright 2012 by ASME

corrosion pit on ID indicates the ID crack was exposed to internal environment (natural gas) for a long period of time.

Figure 27: Crack C Region 3 Fractograph showing fatigue striations Similarly, fracture surface features were also observed in the other two cracks A and B, showing dark stains at ID cracks and fresh surface at OD cracks, confirming that the ID crack formed long ago and the OD crack formed more recently. SUMMARY A well-defined approach was used to identify the root cause for cracking in the dent region. The findings of the root cause analysis are summarized as follows: The careful review of MFL signal characteristics combined with high strain level suggested that the 76% metal loss reported by MFL was not corrosion but associated with cracks or gouge. This prediction was in good agreement with the in-field inspection three cracks originated from the apex of the dent were found in ditch. A shallow dent, 2.7%OD in this case, can be associated with very high strain of 17.2% (ID) and 22.2% (OD), which is nearly three to four times higher than the 6% strain limit as per ASME B31.8. Different dent profiling techniques such as ILI, in-ditch grid and LaserScan consistently showed high levels of strain associated with this dent, implying large external load caused the high plastic strain with possible cracking. The two plastic damage models, namely, DFDI and SLD showed consistently that this dent is susceptible to crack initiation, which demonstrated the effectiveness of the approach and the easy use of simplified DFDI and SLD equations to prioritize the ILI reported dents without performing finite element analysis. The minimum elongation criterion suggests that the dent strain is much higher than the calculated strain limit of 11.5% for in-service, indicating the dent is not acceptable.

Figure 25: Crack C Region 1 Fractograph showing dimples and corrosion pits (SEM) The SEM examination of region 2 (OD surface) showed features of dimples with some corrosion pits (see Figure 26). Again, the OD surface looked fresh and no dark stains observed. These observations confirmed that the OD cracks formed more recently, specifically, during excavation due to re-bounding (pop-out by releasing elastic energy when constrained rock was removed).

Figure 26: Crack C Region 2 SEM fractograph showing dimples The macro examination of the mid-wall or ligament surface (region 3) showed type of beach marks (Figure 24). The SEM examination at this location showed fatigue striations, shown in Figure 27. The leakage of Crack C during the lab pressure cycle test is due to the propagation of ID crack towards OD surface (i.e., breaking the ligament) by fatigue.

11

Copyright 2012 by ASME

Laboratory pressure cycle test demonstrated that this cracked dent failed by leak rather than rupture. Stereoscopic and SEM examination of fracture showed that three distinct fracture regions. The ID fracture surface showed dark stain and dimples with corrosion pits, confirming the dent cracked at the time of denting. The OD cracks were formed more recently during excavation due to a fresh fracture surface with no black stain. The mid-wall fracture surface showed fatigue striation marks formed during the laboratory pressure cycle.

7.

Blade Energy Partners: A Point-to-Point Dent Strain Analysis Tool Using ILI and LaserScan Reported Dent Profile, User Instruction manual, 2009. RTD In-filed inspection Report, March 2010 NACE No.2/SSPC-SP 10 Joint Surface Preparation Standard, Near-white metal blast cleaning (Item no.21066), 2006.

8. 9.

In summary, the investigation showed that this dent was caused by a sharp rock, leading to high strain that caused the partial through-wall cracking at the time of denting during pipeline construction. Since the dent was constrained by a rock, it was survived for a long time. However, further cracking and failure due to fatigue would occur if rock is displaced, for example, due to ground movement. Since the current regulation and the industry guidance require an immediate action on those dents that are associated with crack or high strain level, a further research is highly recommended to set suitable limits and/or damage criteria, which will help pipeline operators to determine the need for remediation or allow for continued operation. REFERENCE 1. 2. 3. BJ ILI Caliper Inspection report, January 2010. ASME B31.8 Gas Transmission and Distribution Piping System (2007). Lukasiewicz, S. A., Czyz, J. A., Sun, C., Adeeb, S. Calculation of Strains in Dents Based on High Resolution In-Line Caliper Survey, IPC2006, Paper No. 10101, 6th International Pipeline Conference, Calgary, Canada, September 25-29, 2006. Gao M et al: Strain-Based Models for Dent Assessment A Review, IPC 2008, Paper No. 64565, 7th International Pipeline Conference, Calgary, Canada Sept. 29 to Oct. 3, 2008. Arumugam U, Tandon S and Gao M et al: Portable LaserScan For In-ditch Profiling and Strain Analysis: Methodology and Application Development, IPC2010, Paper 31336, 8th International Pipeline Conference, Calgary, Canada, September 27-October 1, 2010. Arumugam U, Kendrick D, Limon S, and Gao M et al: An Approach for Evaluating and Prioritizing Dents for Remediation as Reported by ILI Tools, IPC2010, Paper 31401, 8th International Pipeline Conference, Calgary, Canada, September 27-October 1, 2010.

10. J. W. Hancock and A. C. McKenzie, On the mechanisms of ductile failure in high-strength steels subjected to multi-axial stress-states. Journal of Mech. Phys. Solids, Vol. 24, pp 147 to 169, 1976. 11. J. R. Rice and D. M. Tracey, On the ductile enlargement of voids in traxial stress fields. Journal of Mech. Phys. Solids, Vol. 17, pp 201 to 217, 1969. 12. ASME Boiler & Pressure Vessel Code, Section VIII, Division 3, Alternative Rules for Construction of High Pressure Vessels. 2010. 13. ABAQUS V6-11. Reference Manual. 14. Robert B Francini and Nader Yoosef-Ghodsi: Development of a Model for Predicting the Severity of Pipeline Damage Identified by In-Line Inspection Pipeline Research Council International (PRCI) Report, PR-218-063511-B, Final Report No. 08-124, December 2008.

4.

5.

6.

12

Copyright 2012 by ASME

Das könnte Ihnen auch gefallen