UOE Pipes for Ultra Deep Water Application Analytical and FE Collapse Strength Prediction vs. Full- Scale Tests of Thermally Treated Line Pipe

Transcription

1 UOE Pipes for Ultra Deep Water Application Analytical and FE Collapse Strength Prediction vs. Full- Scale Tests of Thermally Treated Line Pipe Andreas Liessem Europipe GmbH Mülheim, Germany Johannes Groß-Weege, Gerhard Knauf, Steffen Zimmermann Salzgitter Mannesmann Forschung GmbH Duisburg, Germany ABSTRACT The present paper presents results of a perennial study on the effect of heat treatment on collapse strength of cold formed pipes, for instance those following the UOE production route. For many years, it has been known that heat treatment, as it is encountered during thermal cycle of pipe coating processes, may compensate the reduction of compressive yield strength owing to cold plastic deformation. The underlying investigation aimed at a systematic analysis of this effect. In particular, it will be shown that appropriate thermal treatment manifests itself positively with respect to collapse resistance, leading to improved pressures. Based on a thorough understanding of the phenomenon, a concept was developed, which is capable of monitoring the positive effect of thermal cycle by appropriate laboratory simulations on compression specimens. This will make it possible to demonstrate improved collapse pressures also under the conditions of serial production based on heat treatment simulations on samples rather than relying on full-scale testing. KEY WORDS: Collapse resistance; heat treatment; plastic collapse; compression strength; heat treatment simulation; quality management INTRODUCTION Advances in technologies for exploration and production do allow previously uneconomic hydrocarbon reserves to be accessed. It is for this reason that future pipelines are being planned to run across difficult environments (Martin, 26), for instance ultra-deep water. This situation naturally entails demanding material and line pipe design issues to be resolved. Line pipe intended for ultra-deep water application predominantly has to be designed with regard to ambient hydraulic pressure action in order to safely withstand plastic collapse. This is due to the fact that off-shore line pipe is not subjected to internal pressure until the pipe string will have been completely installed on the seabed. Design against external pressure is very much different to design against pressure containment. In the first case, stability is the driving parameter governing success or failure and, thus, it is compressive yield strength in circumferential direction rather than tensile yield strength that is to be considered. Hence, the minimum wall thickness requirement is now linked to external pressure. In addition to compressive yield strength, diameter to thickness (D/t) ratio, roundness of cross section (ovality), residual stresses and pipe stress-strain behavior do play a major role. Cold forming operations according to UOE pipe manufacture and subsequent anti-corrosion coating may exert significant effect on characteristic stress-strain behavior of parent material. For instance, the last cold forming operation within the UOE pipe production route, the so-called expansion, would effectuate some reduction of collapse strength if the pipe were not coated. This phenomenon is to be attributed to the Bauschinger effect (1886), which signifies a reduction of yield strength whenever there is a load reversal after previous plastic straining. Today, off-shore line pipe is generally traded with anticorrosion coating. It is for this particular reason that, generally, this effect is not apparent any more. Thermal ageing raises compressive yield strength and, thus, heals out the drop of compressive yield strength caused by cold expansion. From the above, it appears that it is reasonable to account for all relevant production steps in line pipe design. The present paper addresses this issue, notably in the context of thermal treatment of line pipe such as encountered in pipe coating processes. In particular, it will be shown that the adverse effect of cold plastic deformation on compressive yield strength of UOE pipes and, hence, on their collapse strength can be conveniently compensated for by thermal ageing at appropriate temperatures (2º C up to 22º C). This effect has been analyzed and documented in many articles, for instance in Al-Sharif and Preston (1996), DeGeer and Zimmerman (24) and more recently in Tsuru et al. (26) and Liessem et al. (26). Standards intended for off-shore design do neither account for the positive effect of heat treatment nor do they refer to compressive yield strength although it would be more reasonable from a physical point of view. Rather, these resort to tensile yield strength only. For example, DNV-OS-F11 (2) advises to determine a fabrication factor fab for pipes manufactured within a fabrication process that introduces cold deformation (giving different strength in tension and compression). The fabrication factor is then to be applied to the characteristic (tensile) yield stress. It describes the reduction of compressive yield strength relative to specified yield strength. Unless other information exists fab shall be taken as.85 for UOE pipes. However, having in mind the positive effect of thermal heat treatment, DNV-OS-F11 allows in agreement with the customer for an Paper No. ISOPE-27-SBD14 Liessem Total number of pages 8

2 adjustment of the fabrication factor (clause 66, Section 5). Still, public authorities and operators need to be convinced of the effectiveness of thermal treatment of off-shore line pipe. This is the prime intention of the underlying work. The present paper picks up the idea of adjusting fab. In particular, it summarizes the results of a research campaign conducted by Europipe GmbH and its partner Salzgitter Mannesmann Forschung GmbH (SZMF) which was guided by the following methodology: Execution of compression tests on samples of heat treated (HT) and non heat treated (NHT) pipes Heat treatment simulations on compression specimens (lab simulations). This is to prove that the effect of pipe heat treatment can be verified by appropriate material level tests. This is particularly important for quality assurance to be maintained under production conditions. Comparison of results from compression testing and specified minimum yield strength delivers the modified fabrication factor Laying down a quality management procedure that is capable of warranting reproducibility of heat treatment procedure The paper is subdivided in three parts. In the first part, the paper aims at clarifying and quantifying the effect of factors of influence with respect to collapse pressure based on experimental evidence. In the second part of the paper, the results of numerical simulations will be presented and compared to experimental data. This way, it was possible to corroborate observations made in experiments and to reveal parameters of concern, like specimen geometry, sampling position and stress-strain behavior to name a few. In particular, decisions had to be made with respect to the choice of compression specimen geometry in order to accurately capture the positive effect of thermal treatment. It will be demonstrated that extrapolation of results from well-defined laboratory tests to component behavior is possible with a high level of confidence. Finally, it will be concluded that the fabrication factor stated in the offshore standard DNV-OS-F11 applicable for UOE pipe can be increased on condition that heat treatment was performed in accordance with a quality management procedure. For implementation of the findings made in the framework of this research, it must be insured that material properties found in laboratory simulations of thermal treatment and associated component behavior are compliant. This can be accomplished by appropriate quality management. Quality management will be dealt with in the third part of the paper. PART I. EXPERIMENTAL Analysis of collapse resistance of large diameter pipes against external pressure requires adequate and reliable component testing. Pipe samples have to be subjected to test conditions, which closely simulate the in-service situation. As collapse pressure results may significantly depend on boundary conditions experimental results need to be thoroughly analyzed and compared to existing prediction methods. Full-scale tests were carried out at SZMF using end caps. Thus, the pipes were subjected to compressive axial forces (condition CEP ). The pipes tested are compiled in Table 1, along with geometrical data and state of pipe. Additionally, four collapse tests were performed at C-FER (Canada) on relatively short pipe sections ( 2.5 m) without the presence of axial forces (condition CEO ). This was achieved using a special collapse test device. Collapse pressures measured at C-FER were corrected to end condition CEP using appropriate correction factors obtained from finite element analyses performed at C-FER, see DeGeer et al. (26). Identical pipes were tested in the "non-heat-treated" (NHT) (as fabricated) and "heat-treated" (HT) condition. NHT condition correlates with the non-coated pipe condition, that is to say, the condition in which it is found after the expanding operation. The HT condition was achieved by simulating the coating procedure by means of inductive heating. This simulation was performed in a pipe-coating facility operated by Mülheim Pipe Coatings GmbH under realistic production conditions at temperatures in the range of 2 C to 22 C. Table1. Summary of pipes tested Sample OD wt. Heat treatment Grade No. (inch) (mm) none X C air/water X none X C air/water X8 The full-scale collapse tests were accompanied by extensive materials tests, starting with tests on plate material. Samples for investigation of plate properties were first retained from each plate used for pipe production. Two pipes, each of which was 6 m long, were produced from one plate. One pipe per plate was then submitted to coating simulation. From each pipe,.5 m was cut off for further material testing, while the remaining sections of 5.5 m length were tested in the collapse facility subsequent to pipe geometry measurements. API tensile tests were conducted on round bar specimens taken in both longitudinal and transverse direction. Within the circumference, sampling was in two positions. However, the prime emphasis of material testing was laid on compressive behavior of pipe material. For this, round bar specimens of different size were taken from the pipe in circumferential direction in the 3 o clock and 6 o clock position. Both, 67% and 9% of pipe wall thickness was specified as diameter. Specimen length was chosen as twice its diameter. Sampling was as close as possible to the inner diameter. In all cases, two specimens were machined from each location in order to permit checking repeatability of the results obtained. Fig. 1. Sampling of compression specimens The residual stresses of the pipe ring were in all cases determined using the so-called "ring-splitting method". Material property tests Geometrical measurements and ring splitting tests Measurements of outer diameter were performed on pipe ends. Wall thicknesses were also measured at not less than twelve positions on the circumference of the pipe. Customarily, the deviation from the average wall thickness was ±.1 mm for all pipes, see Table 2. Maximum outer OD max and minimum outer diameter OD min as well as the position of maximum diameter measured along the circumference are also reported. Residual stresses were determined using the so-called "ring-splitting method". Here, rings of.5 m in length were split along their axis using saw cuts and the clearance occurring was measured. The residual stresses reported in Table 2 must be interpreted as compressive in the case of the inner surface and as tensile in the case of the outer surface, both being directed in circumferential direction. In

3 principle, the ring-splitting method supposes a linear residual stress profile across the wall. The measurements performed show that heat-treated pipes, in any case, exhibit higher residual stresses than non-heat-treated pipes. Table 2. Results of geometrical measurements and residual stresses Pipe 28"x38 mm X65 as fabricated 28"x38 mm X65 thermally treated 3"x35 mm X8 as fabricated 3"x35 mm X8 thermally treated OD max (mm) OD min (mm) Position OD max o clock o clock o clock o clock t (mm) Residual stress (MPa) Coupon tests In the framework of this research, particular emphasis was laid on establishing a relationship between the results of compression tests performed on material level and the results of full-scale collapse tests. Experience indicates that compression tests on specimens are much more difficult to perform than on tensile specimens. A procedure suitable for conducting compression tests was therefore developed and validated prior to the commencement of the compression tests envisaged within this study. The compression tests performed (Fig. 2) demonstrated that measurements made using strain gages and measurements performed using extensor-meter technique are equivalent in quality. Furthermore, it was demonstrated that even the bending does not have negative effect on compression test results provided that extensometers and gages respectively are mounted oppositely to the test piece. This particular arrangement of instruments insures that the axis about which bending occurs is of little importance. Fig. 3. Compressive yield strength measured on pipe 28 x 38 mm (X65) in the non-heat-treated (NHT) and heat-treated (HT) condition Fig. 4. Compressive yield strength measured on pipe 3 x 35 mm (X8) in the non-heat-treated (NHT) and heat-treated (HT) condition 8 Heat treatment 3 o clock 7 True Stress (MPa) Heat treatment 6 o clock No heat No Heat treatment Treatment 3'o clock 3 o clock No Heat Treatment 6'o clock Heat Treatment 3'o clock No heat treatment 6 o clock Heat Treatment 6'o clock 1 Fig. 2. Testing coupons at SZMF Compression-test specimens of 67% wall thickness and those of 9% wall thickness were tested in all cases. The results of compression tests (.2% offset) are depicted in Figs. 3 and 4. In Fig. 5, stress-strain curves of heat treated and non-heat treated pipe material are shown. The curves illustrate that the Bauschinger effect usually present in cold-worked pipes is significantly relieved by heat treatment.,%,2%,4%,6%,8% 1,% 1,2% 1,4% 1,6% 1,8% 2,% True Strain Fig. 5. Stress-strain curves of pipe 3 x 35 mm (X8) Full scale tests (collapse tests) A new collapse-test device was designed at SZMF (Duisburg) for execution of full-scale collapse tests on large diameter pipes. Essentially, this system consists of a seamless containment (Fig. 6) fabricated by means of a hot forming process. This fabrication process is in favour of the load-bearing capabilities of the system since it does not feature any welds.

4 The vessel has an internal diameter of 85 mm and thus permits testing of pipes with a diameter of up to 32". The maximum pipe length which can be tested is approx. 6.5 m. This ensures that the "pipe body" to be tested can always have a length of not less than eight times the pipe diameter. The collapse-testing system has been designed for a maximum pressure of 7 bars and thus covers testing of longitudinally welded large-diameter linepipes with wall thicknesses of up to 5 mm and above. At maximum pressure, a force corresponding to around 4 tons acts on the vessel top, which absorbs the axial force by means of tie bars. Fig. 8. Record of collapse test Fig. 6. Pressure vessel for collapse testing End caps must be welded on prior to execution of the collapse test (see Fig. 7). Once equipped with end caps, the pipes are inserted into the containment and the vessel is closed by means of abovementioned tie bars. Prior to testing, the vessel is flooded with water and degassed. During the test, internal pressure is continuously increased by means of a constant feeding rate. The volume of water fed and the pressure achieved were continuously recorded using a data acquisition system. Fig. 7. Collapse testing facility at SZMF Fig. 8 shows the record of a test on a 28" x 38 mm (X65) pipe. At the beginning of the test, internal pressure rises linearly with time, since the pressure pump is feeding a constant volume of water into the test vessel. When the maximum pressure is reached the pipe collapses and the internal pressure falls off sharply. This is because an inward bulge, or dent ("buckle"), develops in the pipe (shell buckling). Upon continuation of the test the internal pressure may rise again at a lower level, finally remaining constant for a longer period. This, practically constant pressure, is necessary in order to propagate the dent along the longitudinal axis of the pipe ("buckling propagation pressure"). PART II. NUMERICAL ANALYSES Finite element calculations Fully 3-dimensional analyses using eight-node brick elements were performed in order to predict collapse pressures as accurate as possible. Preliminary analyses have shown that 6 elements over the wall thickness and 12 elements over the circumference were a good compromise in terms of accuracy and calculation time. The pipe material was assumed to undergo isotropic, elastic-plastic material behaviour with Young s modulus and Poisson s ratio equal to 21 MPa and.3 respectively. Stress strain curves measured in uni-axial compression tests on cylindrical specimens, taken in circumferential direction, were used. Two different specimen sizes were investigated. The first one had a diameter of about 67% of pipe wall thickness and the location was chosen as close as possible to the inner surface. The second one had a diameter of about 9% of pipe wall. In both cases, two locations were considered, namely the 3 o clock and the 6 o clock position. The pipe section was modeled based on the measured inner outline, but with a constant wall thickness. Two different pipe end conditions (CEO, CEP) were considered. Condition CEO represents to a pipe with capped ends but without any pressure on caps, i.e. without axial load in the pipe. Condition CEP represents the situation, which was realized in the full scale collapse tests performed at SZMF. The pipe has a capped end which is also subjected to external pressure. Therefore an axial load is acting on the pipe. In all cases two test pipes were considered, the first one was the as fabricated (non-heat treated) pipe, the second one was a thermally treated pipe. In the finite element simulations two variants were investigated with respect to material behavior: 1. For the first variant, the stress-strain curves from 9% wall specimens at 3 o clock and 6 o clock position were applied. Correspondingly, two different stress strain curves around the circumference were adopted but there was no variation over the wall thickness. 2. For the second variant the stress-strain curves from the 67% wall specimens were adopted at 3 o clock and 6 o clock position. The material behavior was again assumed to be constant over the wall thickness. For both variants the analyses were carried out for the end conditions CEO and CEP. Furthermore the influence of residual stresses was investigated. The measured data of the test pipes is summarised in Table 2. Typical stress-strain curves are shown in Fig. 5. In the as fabricated condition a gradual transition from the elastic to the plastic

5 behaviour occurs as a result of the Bauschinger effect. The results of the calculations for both pipe dimensions and both condition (as fabricated and thermally treated) are shown in Figs. 9 to 12. The experimental test result obtained at SZMF is also given in each Fig Pipe 3" od x 35 mm wt, grade X 8, thermally treated Collapse pressure [bar] Pipe 28" od x 38 mm wt, grade X 65, as fabricated % wall specimen 67% wall specimen Specimen size for compression tests CEO - without residal stresses CEO - with residual stresses CEP - without residual stresses CEP - with residual stresses Experimental collapse test Fig. 9 Results for pipe 28 x 38. mm (X65), as fabricated Collapse pressure [bar] Pipe 28" od x 38 mm wt, grade X 65, thermally treated CEO - without residal stresses 15 CEO - with residual stresses CEP - without residual stresses CEP - with residual stresses 1 Experimental collapse test 5 9% wall specimen 67% wall specimen Specimen size for compression tests Fig. 1 Results for pipe 28 x 38. mm (X65), thermally treated Collapse pressure [bar] Pipe 3" od x 35 mm wt, grade X 8, as fabricated CEO - without residal stresses 15 CEO - with residual stresses CEP - without residual stresses 1 5 9% wall specimen 67% wall specimen Specimen size for compression tests CEP - with residual stresses Experimental collapse test Fig. 11 Results for pipe 3 x 35. mm (X8), as fabricated Collapse pressure [bar] Fig. 12 9% wall specimen 67% wall specimen Specimen size for compression tests CEO - without residal stresses CEO - with residual stresses CEP - without residual stresses CEP - with residual stresses Experimental collapse test Results pipe 3 x 35. mm (X8), thermally treated Conclusions Following conclusions can be drawn from the FEM calculations: Test pipes in as fabricated condition (Figs. 9 and 11) The difference between the results for the 67% wall specimens and the 9% wall specimens were very low (less than 3%). The 9% wall specimens gave higher collapse pressures for the pipe 28 x 38. mm and lower pressures for the pipe 3 x 35. mm. The collapse pressures by FEM for condition CEP were lower than the test result (test condition CEP). The error was between 5% and 1%. The influence of the axial pressure on the end cap was relatively low (less than 4%). The influence of the residual stresses was very low (about 1% maximum). In most cases the residual stresses lead to a reduction of the collapse pressure. Test pipes in thermally treated condition (Figs. 1 and 12) Like in the case of the as fabricated pipes the differences in the collapse pressures between the 9% wall specimens and the 67% wall specimens was very low. The 9% wall specimens gave slightly higher collapse pressures than the 67% wall specimens. For test pipe 28 x 38 mm the collapse pressures by FEM for condition CEP (with and without residual stresses) were lower than the test result. The error was about 13%. For test pipe 3 x 35. mm the collapse pressures by FEM without residual stresses were higher than the test result. With residual stresses the collapse pressures were lower. For test pipe 28 x 38. mm the influence of the axial pressure was between 5% and 7%, always leading to higher collapse pressures. For test pipe 3 x 35. mm the influence of the axial pressure was relatively low. The influence of the residual stresses was much higher than for the test pipes in the as fabricated conditions. The difference was between 4% and 6% (for 28 x 38. mm) and between 6% and 14% (3 x 35. mm). In all cases the influence of residual stresses gave lower collapse pressures. Comparison with experimental results The ratio of the collapse pressures predicted by FEM to the collapse pressures in the tests is shown in Fig. 13 for all the calculation variants and all test pipes. Concerning the finite element results only end condition CEP has been considered here, because this condition corresponds to the condition in the collapse tests. The results were drawn over the ratio of the elastic collapse pressure P e to the plastic collapse pressure P p, where P e and P p are defined according to the DNV formula

6 with Pressure (FEM) / Pressure (collapse test) [-] Fig. 13 P e 2 E ( t / D) 2 (1 ) Pp 2 f y fab t / D 3 E Young s modulus (21 MPa) v Poisson s ratio (.3) t Wall thickness D Outer diameter f y Yield strength (SMYS) fab Fabrication factor =.85 non-heat treated condition (NHT) = 1. heat treated condition (HT) 3" x 35 mm, X8, HT Comparison of collapse pressures FEM (for CEP) versus collapse test 3" x 35 mm, X8, NHT P e < P p ("Elastic" collapse) P e > P p ("Plastic" collapse) P e / P p - ratio [-] Comparison of collapse pressures 67% wall specimens - without residual stresses 67% wall specimens - with residual stresses 9% wall specimens - without residual stresses 9% wall specimens - with residual stresses 28" x 38 mm, X65, HT 28" x 38 mm, X65, NHT It can be observed that for P e > P p, i.e. for plastic collapse modes the predicted collapse pressures for all variants were lower than the test result. In this case the FEM obviously gives conservative predictions. The maximum error is about 12%. On the other hand, for the case P e < P p, i.e. for elastic collapse modes the influence of residual stresses needs to be taken into account in order to obtain a conservative prediction. Concerning the influence of the specimen size it can be observed, that the difference between the results for the 67% wall specimens and the 9% wall specimens is very small. In all cases the influence of the specimen size was lower compared to the influence of the residual stresses, in particular for the heat-treated condition. The influence of the axial pressure on the end cap can be determined from the finite element analyses. Fig. 14 shows the ratio of the predicted collapse pressures with axial pressure (CEP) to the predicted pressures without axial pressure (CEO) for all the calculation variants and all test pipes. The results are drawn over the D / T ratio. In general the axial pressure leads to higher collapse pressures. The maximum difference is about 7%. There is also a clear influence of the thermal treatment. On the other hand, the influence of the residual stresses is very low. In order to compare the different calculation variants concerning their prediction accuracy the mean values and standard deviations of the ratios of the predicted collapse pressures to the test results was determined for all test pipes. The results are shown in Fig. 15. It can be seen that 67% and 9% wall specimen gave conservative results. Taking into account also residual stresses the standard deviation was very low. FEM (with p axial) / FEM (without paxial) [-] Fig. 14 Predicted collapse pressure / collapse test result [-] 28" x 38 mm, X65, HT 28" x 38 mm, X65, NHT Influence of axial pressure on end cap FEM results 3" x 35 mm, X8, HT + NHT Fig. 15 D / T - ratio [-] Influence of axial pressure on end cap Comparison of prediction methods Prediction method: 1a - FEM, CEP, 67% wall specimen, without residual stress 1b - FEM, CEP, 67% wall specimen, with residual stress 2a - FEM, CEP, 9% wall specimen, without residual stress 2b - FEM. CEP, 9% wall specimen, with residual stress 67% wall specimens - without residual stresses 67% wall specimens - with residual stresses 9% wall specimens - without residual stresses 9% wall specimens - with residual stresses 1a 1b 2a 2b Prediction method Comparison of prediction methods Mean value Standard deviation DNV prediction versus collapse test results The results of all the collapse tests performed at SZMF and at C-FER have been compared with the DNV prediction. In particular, different yield strengths have been introduced in the DNV formula: Tensile yield strength measured on round specimens in circumferential direction at.2% plastic offset or at.5% total strain. Specified minimum yield strength (SMYS). Compressive yield strength measured on large column specimens in circumferential direction at.2% plastic offset. The specimen sizes were 67% and 9% of wall thickness. In all cases the specimens have been taken both at 3 and 6 o clock position of the pipes. For the measured tensile yield strength as well as for the SMYS the fabrication factor was applied (.85 for pipes as fabricated and 1. for thermally treated pipes). The results of the statistical evaluations are shown in Fig. 16 for the as fabricated pipes and in Fig. 17 for the thermally treated pipes. In particular, the mean values of the ratios of the DNV prediction to the collapse test results were evaluated. Following conclusions can be drawn from the evaluations: The DNV prediction using measured tensile yield strengths values in combination with the fabrication factor under-estimates the collapse pressures in all cases. The error is between 11% and 14%. The 3 o clock position gave slightly higher collapse pressures than the 6 o clock position.

7 The DNV prediction using the SMYS in combination with the fabrication factor also over-estimates the collapse pressures, although the error is very small (2-3%). Using.2% plastic offset compressive yield strength the DNV prediction estimates well collapse pressures of pipes in the as fabricated condition. The error is 5-8%. Using.1% plastic offset would be over-conservative. The error is about 14%. There is neither significant difference between the 3 and 6 o clock position nor between specimen geometries (67% and 9% wall thickness) For the thermally treated pipes the result is more or less similar, although the errors are much smaller. Thus, collapse pressures of thermally treated pipes can be predicted well using the DNV equation on condition that: Measured compressive yield strength is used Compression yield strength is defined as.1% plastic offset Sampling is in the 6 or 3 o clock position Specimen size is equal to 9% or 67% of wall thickness Ratio DNV prediction / collapse test result [-] 1,1 1,5 1,,95,9,85,8 Fig. 16 Ratio DNV prediction / collapse test result [-] 1,2 1,15 1,1 1,5 1,,95,9,85,8 Fig. 17 Collapse pressure predition by DNV formula for as fabricated pipes.2% offset.5% total Measured ovality SMYS d/t=9%.2% offset Variant d/t=67%.2% offset Mean values, 3 o'clock Mean values, 6 o'clock d/t=9%.1% offset Collapse pressure prediction by DNV as received pipes Collapse pressure predition by DNV formula for thermally treated pipes.2% offset.5% total Measured ovality SMYS d/t=9%.2% offset Variant d/t=67%.2% offset Mean values, 3 o'clock Mean values, 6 o'clock d/t=9%.1% offset d/t=67%.1% offset d/t=67%.1% offset Collapse pressure prediction by DNV thermally treated pipes PART III. QUALITY MANAGEMENT PROCEDURE Background Heat treatment parameters such as time and temperature must be described with sufficient accuracy. In order to correlate the asfabricated and the coated condition, it is necessary to perform lab simulations of heat treatment on compression specimens and to compare their yield strength to this one of non-heat treated compression samples. For the purpose of precisely representing heat treatment caused by coating, measurement of the temperature distribution during pipe coating had to be executed first. This was done at Mülheim Pipe Coatings GmbH under the premise of adhering to genuine production conditions. In parallel, various simulations of this heat treatment were performed on coupons, which were taken from the same pipe prior to coating. The results of these tests were then assessed against the results obtained from tests performed on coated pipe. Temperature distribution within typical coating process For the sake of measuring the temperature distribution during a pipe coating process, a suitable pipe was selected from ongoing production. A ring was cut off, which was then used for machining various specimens. One set of specimens was used for characterization of the as fabricated state. Another set was employed for heat treatment simulations. A data logger protected by an insulation box was fixed inside the pipe. In so doing, it was possible to record the temperature distribution along circumference and wall thickness over time. In radial direction, three thermo-couples were used. One was fixed closely to the inner surface, one at the center and another closely to the outer surface. Temperatures were recorded throughout the complete coating process. Fig. 18 shows a representative plot of the heating and cooling sequence. In a pre-heating sequence, pipes are pre-warmed up to a moderate temperature (area 1 in Fig. 18) within the feeding line. Then, prior to applying adhesives and actual coating, the pipe is heated up by means of two induction coils (areas 2 and 3 in Fig. 18) aligned in a row. Due to the high heat capacity of steel the loss of temperature owing to convection is comparatively low prior to active cooling (water). This period of time in between end of second induction coil and active cooling is defined as the holding time (area 4). Along the circumference, temperature records were very much the same. Temperature ( C) Middle of wall Outside wall Inside wall Time (min.) Fig. 18. Temperature against time during pipe coating M2, 14 mm M3, 4 mm M4, 27 mm Laboratory simulation of heat treatment In order to comply as close as possible with the situation in the pipe mill heat treatment simulations in the laboratory were executed using salt baths. This strategy permits heating as rapid as encountered in inductive heating processes. Compression testing was performed on a variety of specimen geometries. Like in the accompanying lab tests of collapse experiments (see Part I), column specimens (squared as well as round) were machined as close as possible to the inner surface in order to capture the location where circumferential compressive stress is greatest. The results from laboratory simulations of heat treatment confirmed the following: Heat treatment can be well simulated using either squared or round specimens Heat treatment lifts up compressive yield strength significantly

8 Decreasing or increasing heat treatment time with respect to reference time does not cause significant changes of compressive yield strength Quality management Preliminaries A specification has been drafted, which is additional to costumer specifications. Hence, the procedures described below are to be understood as amendments to those required by the customer with the specific intent to qualify off-shore line pipes with the label ultracollapse resistant. The procedures, hereafter referred to as Manufacturing Procedure Qualification Tests (MPQT), shall be applied in conjunction with DNV-OS-F11. Manufacturing Procedure Qualification Tests Manufacturing qualification tests should be performed ahead of production in order to allow for evaluation. It is desirable to align details of any specific MPQT in agreement with the customer. Once qualification has been awarded exactly the same process parameters have to be maintained during production as assumed during qualification. Any variation must be subjected to re-qualification. A minimum number of pipes shall be manufactured from different heats. Per heat, a limited number of pipes should be selected for MPQT, in addition to all inspections usually required during production. The result of all tests will be evaluated by the client. The mill will have to be authorized by the client to produce on condition that all requirements of the MPQT have been fulfilled. Central to qualification are the following items: The manufacturer defines a heat treatment procedure. Particularly he proves that heat treatment parameters are maintained constant during production. The manufacturer shall perform lab heat treatment simulations on compression samples. The parameters employed shall comply with those of the heat treatment procedure. In particular, parameters are to be deduced conservatively from the temperature vs. time history graph (Fig. 18) of heat treatment. Simulations can be performed using either salt bath or any other approved method, for example induction heating. Compression strength shall be determined on column samples machined in transverse direction in the 3 and/or 6 o clock position, see Fig. 19. The cross section of the samples may be squared or round. Diameter and edge length respectively shall make up not less than 2/3 of the wall thickness. The compressive yield strength is defined at preferably.2% plastic offset strain Compressive yield strength of samples from heat treated pipe and samples from lab simulations shall show clear correlation. The ratio of compressive yield strength upraised on samples from pipe and SMYS delivers adjusted fabrication factors During serial production the positive effect of heat treatment shall be continuously reviewed and verified by means of ongoing laboratory simulations of thermal cycle and subsequent compression testing. Always the lowest value specified within the MPQT should be achieved. 6 Fig. 19. Definition of sampling positions 3 CONCLUSIONS Within this paper, aspects of thermal cycle of anti-corrosion coating were discussed with respect to collapse resistance of UOE pipes. Central to the discussion was the effect of heat treatment on compressive yield strength in transverse direction, lifting up collapse pressure to levels, which are significantly higher than predicted by DNV-OS-F11. Issues like numerical assessment of collapse pressure of UOE pipes taking into account heat treatment effects were dealt with. Also, implications with experimental testing like sampling procedure and specimen geometry were addressed. A way was described upon which practical implementation of the knowledge acquired into pipe production could be achieved. It is based on the following items: Compression testing on samples that were subjected to laboratory heat treatment simulations rather than on specimens gained from coated pipe The cross section of compression specimens may be squared or round. Diameter and edge length respectively shall not be less than 2/3 of pipe wall thickness Compression yield strength is defined as the stress corresponding to a.2 plastic offset strain Abovementioned items are laid down in a quality management procedure paving the way for implementation into practice. REFERENCES Al-Sharif, A.M. and Preston, R Improvement in UOE Pipe Collapse Resistance by Thermal Aging. Proceedings of the 28th Annual Offshore Technology Conference, OTC 8211, Houston, May, pp DeGeer, D.D. and Zimmerman, T.J.E Testing Deepwater Pipelines. Proc of the Deepwater Pipeline Technology Conference, Clarion Technical Conferences and Pipes & Pipelines International, New Orleans, Louisiana, March DeGeer D., Vitali, L., Marewski U., Hillenbrand H.G., Weber B., Crawford M. 24. Collapse testing of thermally treated line pipe for ultra-deepwater applications. Proc of OMAE4 23rd International Conference on Offshore Mechanics and Arctic Engineering, June 2-25, Vancouver, Canada DeGeer, D., Piers, K., Timms, C., Xie, J., Tsuru, E., 26. Collapse testing short linepipe for deepwater appclications. Proc of ISOPE Graef, M.K., Marewski, U., Vogt, G Collapse Behavior of Offshore Line Pipe under external Pressure. 9th Symposium on Pipeline Research, September 3-October 2. Houston Texas Hillenbrand, H.G., Graef, M.K., Groß-Weege, J., Knauf, G., Marewski, U. 22. Development of Line Pipe for deep-water applications. Proc. of 12th International Offshore and Polar Engineering Conference & Exhibition. Kyushu Kitakyushu. Japan. May Liessem, A., Marewski, U., Groß-Weege, J., Knauf, G. 26. Methods for Collapse Prediction of UOE Line Pipe. Proc 25 th International Conference on Offshore Mechanics and Artic Engineering, Hamburg, Germany Martin, R 26. "Investment in Oil and Gas Filed Materials Technology - The Key to a Secure Energy Supply. Energy Materials, Vol 1, pp Tsuru, E., Asahi, H. (26): Improved Collapse Pressure of UOE Line Pipe with Thermal Aging for Deep Water Applications, Proc of 16 th International Offshore and Polar Engineering Conference (ISOPE), San Francisco, Vol. 4, pp Stark, P.R. and McKeehan, D.S Hydrostatic Collapse Research in Support of the Oman India Gas Pipeline. Proceedings of the 27th Annual Offshore Technology Conference, OTC 775, Houston, May, pp

9 DNV 2. DNV OS-F11, Submarine Pipeline Systems. Det Norske Veritas Kyriakides 199. Factors affecting Pipe Collapse. Seminar on Collapse of Offshore Pipelines. Houston Texas. Gresnigt, van Foeken, Shilin Chen 2. Collapse of UOE Manufactured Steel Pipes. Proceedings of the Tenth international Offshore and Polar Engineering Conference. Seattle, USA, May 28 June 2. Torselletti E., Vitali L., Bruschi R., Collberg L., 23. Minimum Wall Thickness Requirements for Ultra Deep-Water Pipelines. Proceedings of OMAE3 22nd International Conference on Offshore Mechanics and Arctic Engineering. June 8-13, Cancun, Mexico