Technology Burst Test, Finite Element Analysis and Structural Integrity of Pipeline System By Nasir Shafiq, Mokhtar Che Ismail, Chanyalew Taye, Saravanan K and M F Nuruddin Abstract Pipelines are one of the major means of transporting hydrocarbons (oil and/or natural) from one point to the other point, which may be routed within onshore or offshore locations. There is a great risk to occur many defects during the service life these pipelines. Corrosion (internal and/or external) is one of the common defects observed in many instants. With the passage of time the corrosion that occurs either at a localized point or onto a large area cause the metal loss hence the strength or in other words load bearing capacity of the pipeline is reduced. In order to continuously operate the pipeline for suiting the earlier assumptions taken at the time of first design, it requires strengthening for returning to the original capacity. The type of strengthening technique and procedure required to estimate the percentage capacity lost using appropriate experimental procedure. Full-scale pipe burst test is most commonly used for this reason. A systematic study has been conducted for remaining capacity analysis of API 5L X52 pipeline operated and managed in the South China Sea. The objective of the study was to investigate the fracture behaviour of corroded pipe extracted from in-service pipeline using full scale burst test and to correlate the experimental results with the finite element analysis. INTRODUCTION Pipeline is exposed to both internal and external corrosion. Internally, pipeline carrying hydrocarbon fluid is known to be susceptible to CO 2 corrosion. Externally, harsh marine environment becomes a threat if the preventive corrosion measures fail. The CO 2 corrosion is not only influenced by the presence of the corrosive species in the fluid such as H 2 S and organic acid but also enhanced by operational parameters such as temperature, ph, flow rate, water cut and others. The presence of Sulfate reducing bacteria (SRB) poses the pipe to another aggressive corrosion; microbial induced corrosion (MIC) 1. As part of the corrosion management, inspection and monitoring relies on intelligent pig (IP) data obtained from the scheduled pigging exercise to provide feedback on the integrity of the pipeline. The IP metal loss data provided a means for engineers to conduct fitness for service assessment based on the available codes such as DNV etc. Since the FFS assessment depends on the accuracy of the IP data, it is of upmost importance to have high confidence level of the IP data. Overestimation of metal loss by IP means premature retirement of the pipe. Furthermore, the codes used also contain inherent conservativeness which also affects the assessment. Thus, there is a need to establish the understanding of the accuracy of the IP data and also the degree of the conservativeness of the codes used. Recommendation based on burst test investigation of naturally corroded pipeline is considered to be an appropriate technique for improving the conservativeness of current assessment procedures. The main objective of this project is the failure analysis of PMO PL24 corroded pipeline. Burst test of pipe removed from service containing natural corrosion defects and simulated defects were conducted. Furthermore, nonlinear finite element analyses of these defects were conducted 2. A 10 diameter pipeline of 6.9 km length carrying wet 38 JUL-SEPT 2010 Visit our websites at www.safan.com / www.pm-pipeliner.safan.com
and semi processed crude oil from platform-a platform-b is operating in South-China Sea. It was commissioned in 1982 with a design life of 20 years. Maximum Allowable Operating Pressure (MAOP) was 40.0 bars, which was derated pressure from 93 bars based on the Fitness-for- Service (FFS) Assessment performed by the operator in 2005. The pipeline had been operated at an average Operation Pressure (OP) of 28.0 bars by the time when the operator/consultant referred to the University Technology PETRONAS. An inline inspection using Magnetic Flux Leakage (MFL) tool was performed in November 2006, it reported 10,804 metal loss defects; where 10,803 internal defects concentrated at 700 metres from platform-a. There was only one external defect reported at the riser of the Platform- A. Based on the fitness for service, FFS done using DNV-RP- F101Part A; 17 defects were reported with allowable corroded pipe pressure (P corr ) lower than the maximum allowable operating pressure MAOP and at the location of 8 defects, the P corr was determined as zero. Furthermore, there were 10 groups of interacting defects having P corr = 0. All the identified defects were located within 100 to 350 metres section from platform-a. Accordingly, it was recommended to replace critical section from log distance 93 to 850 metres due to single and interacting defects having P corr = 0 bar and containing highest density of defects (88%). Universiti Teknologi PETRONAS (UTP) is commissioned to perform failure analysis on this retired section of the pipeline. For the study purpose, about 100 meter of the retired pipe section was delivered to UTP. Objectives and Scope The main objective of this study was to establish the relationship between the maximum allowable pressure, MAOP due to corrosion defect reported by the IP and the burst pressure based for stress analysis of the pipe containing natural and artificial corrosion defects and through the nonlinear finite element analysis of known defect profiles. The sub-objectives include: Obtain the cross-sectional profile of the corroded pipelines Verify the reliability of IP tally by the UT-Scan results Experimental investigation of the burst pressure for the corroded pipelines Evaluation of defect assessment methods, by comparison of burst test results with predicted results for pipeline with corrosion defects, Comparison and verification of burst pressure with finite element simulation. Develop a model to predict the burst pressure of pipelines with longitudinal defect under internal pressure Methodology The study was conducted according to the following framework. Visual Inspection: Conduct visual inspection and take photographic records of visible defects, Perform UT-Scan and record the profile of the identified sections FFS Calculation: FFS calculations for 10 critical sections identified by IP inspection by ASME B31G, modified ASME B31G and DNV codes Material Test: Cut tensile test specimens from the pipeline and prepare the surface according to the standard (ASTM E 8M 04). Conduct tensile test and record the necessary data to be used for burst strength prediction Burst Test: To identify critical sections for performing the burst test based on UT-Scan Cut the test samples and clean the surface weld the end cup, fix the strain gauges, attach pressure transducer gauge, and attach water inlet perform burst tests and record the data Finite Element Simulation Non-linear FE analysis was performed using ANSYS software for simulating the experimental results obtained by the burst test. Burst Pressure (P f ) Predictions Based On IP Data A summary of the critical metal loss sorted in order of their absolute distance from the launch is shown in Table- 1, these sections were reported for ERF value greater than 1. Upon the decision made by the operator, the 450 m of critically corroded section was replaced with the new section by contractor from17 to 24 of Aug 2008. For these identified features, burst pressure predictions were made by using ASME B31G, modified ASME B31G, DNV-RP-F101Part A and FEM. The summary of the results are given in the Figure-1 and 2. The following values were used for the calculations. Pipe size: D=273.05mm, t= 11.1mm D= 274mm, t=12.8mm Yield strength: SMYS, (actual, engineering) = 358 MPa Tensile strength: σ UTS (actual, engineering) = 455 MPa Sample Calculation (D=273.05 mm, t = 11.1mm): Let us consider defect designated by Feature Nr8204 to demonstrate the failure pressure prediction by using the above mentioned methods. The defect geometries for this feature are: d/t = 0.41 (41%) and L m = 250 mm. JUL-SEPT 2010 39
Table-1: Summary of inspection sheets with the metal loss features sorted in order of their absolute distance from the launch ASME B31G First let us calculate A and P Therefore, according to modified ASME B31G, the burst (failure) pressure of the pipe due to defect identified as Nr8204 is 251 bar. For A is greater than 4.0: DNV-RP-F101 burst capacity First let us calculate factor Q, Therefore, according to ASME B31G, the burst (failure) pressure of the pipe due to defect identified as Nr8204 is 188.9 bar. Therefore, the failure pressure (P f ) is calculated using: Modified ASME B31G First let us calculate the Folias factor, M Therefore, according to DNV RP-F101 the burst capacity of the pipe due to defect identified as Nr8204 is 281.3 bar. 40 JUL-SEPT 2010 Visit our websites at www.safan.com / www.pm-pipeliner.safan.com
Sample Calculation (D= 274 mm, t = 12.8 mm) : Let us consider the same defect as identified by previous sample calculation. DNV-RP-F101 burst capacity First let us calculate factor Q, ASME B31G First let us calculate A and P Therefore, the failure pressure (P f ) is calculated using: For A is less than 4.0: Therefore, according to DNV RP-F101 the burst capacity of the pipe due to defect identified as Nr8204 is 329.1 bar. Therefore, according to ASME B31G, the burst (failure) pressure of the pipe due to defect identified as Nr8204 is 287.5 bar. Modified ASME B31G First let us calculate the Folias factor, M Therefore, according to modified ASME B31G, the burst (failure) pressure of the pipe due to defect identified as Nr8204 is 290.4 bar. MAOP (P corr ) Predictions Based On IP Data The maximum allowable operating pressure or the allowable corroded pipe pressure can be estimated based on operator s factor of safety. In the case of ASME B31G codes and the FE calculation, MAOP can be calculated by considering an appropriate design factor (F). We may use JUL-SEPT 2010 41
the acceptance equation of DNV-RP-F101Part A for direct calculation of P corr. The allowable corroded pipe pressure of a single metal loss defect subject to internal pressure loading is given by the following acceptance equation. The partial safety factor for MFL under normal safety class can be taken as, γ m = 0.74. For 80% confidence level the standard deviation can be taken as, StD[d/t] = 0.08, therefore, γ d = 1.28 and ε d = 1.0 can be taken. Let us assume safety factor, (F = 0.5) to recalculate the acceptable operation pressure for ASME codes and FE simulations. Burst Test Results As per the critical section identified by UT P-scan, five burst test samples were prepared from section A and section B. One of these test samples (T2) is with general corrosion distributed over the internal surface and four of which were with artificial longitudinal corrosion defect. The defects were smooth and machined on the external surface of the pipe. The simulated corrosion defects were machined on at the external surface. The summary of the tests are shown in Table-3. The defects had a smooth surface and all edges were made with a small radius. The surfaces were also grinded slightly to get smoother surface for ease of thickness measurements and attachment of strain gauges. At each end of the test specimens a spherical end caps were welded to the pipe. The inlet for internal pressure and mounting of the pressure transducer were at the end flanges. Water was pumped into the pipes until burst. The water pump used had a capacity of 1000 bar. All tests were performed at room temperature of about 30 o C. Table 2: MAOP (P corr ) predictions 42 JUL-SEPT 2010 Visit our websites at www.safan.com / www.pm-pipeliner.safan.com
Table-3: Description of test samples * Sample failed during machining of for 85% simulated defect. Burst tests were successfully conducted for test sample T001 to T004, but test sample T005 failed during machining of simulated defect due to imbedded pinhole defect on the corroded pipeline internal wall. The corresponding numerical simulation results were also calculated. FE simulations were used for simulated defects and analytical calculation was made for test sample T002. This test sample (T002) was with general corrosion defect evenly distributed over the internal surface of the pipeline. Therefore, in order to make analytical calculations, we may assume uniform wall thickness reduction due to the general corrosion. According to the P-Scan measurement in this section the average wall thickness of this test sample is 10.58 mm, only about 5% wall loss. We may use the maximum hoop stress theory to predict the burst pressure of this section. The maximum hoop stress theory suggests the maximum allowable pressure for hollow pressure vessel (cylinder) as follows. Therefore, the failure pressure predicted is about 368 bars, which is within 4.4% deviation of the experimental burst pressure. Similarly, as shown in Table-3, the errors for FE simulations for T001 and T003 were 3% and 3.2%, respectively. CONCLUSIONS Based on the visual inspection, corrosion is found to be minimal and concentrated in the bottom half of the pipeline between 3 to 9 o clock positions. There was no severe localized and deep corrosion of any form. Rusting and sea water infiltration under the newly wrapped rubber laminations was observed. Therefore, proper lamination should be done after inspection or maintenance activities. Among the three codes used, DNV gave less conservative burst strength predictions. The burst test at the most severely corroded section of the pipeline, as it is represented by test sample T002, was 385 bars. This pressure is much higher than the design pressure of the pipeline. Therefore, it can be concluded that the level of corrosion in this pipeline is not significant to challenge the integrity. FE simulation results estimated the burst capacity of the pipe within reasonable accuracy less than 5%. PP References 1. O.H. Bjornoy and M.D. Marley. Assessment of Corroded Pipelines: Past, Present and Future, Proceedings of tile Eleventh (2001) International Offshore and Polar Engineering Conference Stavanger, Norway, June 17-22. 2. Duan Qingquan, Zhang Hong, Yan Feng, Deng Changyi. Hydrostatic burst test 0F X80 grade steel pipe, Journal of Loss Prevention in the Process Industries, (2009) pp: 897 900. PetroMin Pipeliner would like to show appreciation to Prof Nasir Shafiq and his faculty in Universiti Tecknologi PETRONAS, for this detailed article which was presented at our Hydrocarbon Asia s Engineering and Operations Convention, 20 21 July 2010, in Kuala Lumpur, Malaysia. JUL-SEPT 2010 43