Failure analysis of corrosion at an inhomogeneous welded joint in a natural gas gathering pipeline considering the combined action of multiple factors

    Failure analysis of corrosion at an inhomogeneous welded joint in a natural gas gathering pipeline considering the combined action of multiple factors Qiao Qiao, Guangxu Cheng, Wei Wu, Yun Li, Hui Huang, Zefeng Wei PII: DOI: Reference:

S1350-6307(16)30040-1 doi: 10.1016/j.engfailanal.2016.02.015 EFA 2823

To appear in: Received date: Revised date: Accepted date:

7 August 2015 13 January 2016 11 February 2016

Please cite this article as: Qiao Qiao, Cheng Guangxu, Wu Wei, Li Yun, Huang Hui, Wei Zefeng, Failure analysis of corrosion at an inhomogeneous welded joint in a natural gas gathering pipeline considering the combined action of multiple factors, (2016), doi: 10.1016/j.engfailanal.2016.02.015

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Failure analysis of corrosion at an inhomogeneous welded joint in a natural gas gathering pipeline considering the combined action of multiple factors Qiao Qiao a, Guangxu Cheng a,, Wei Wu a, Yun Li a, Hui Huang b, Zefeng Wei a School of Chemical Engineering and Technology, Xi’an Jiaotong University, Xi’an 710049, PR China China Special Equipment Inspection and Research Institute, Beijing 100013, PR China

T

b

IP

a

SC R

Abstract

AC

CE P

TE

D

MA

NU

A gathering pipeline in a gas field in Northeast China transporting mainly natural gas but also some water and sands was welded using two 20G pipes with different wall thicknesses. The internal corrosion of this pipeline was inhomogeneous. The unequal wall thickness welded joint suffered more serious corrosion damage than did other parts of the pipe and was found to fail during routine maintenance. In this paper, the corrosion damage at this joint was studied based on inhomogeneity. Internal and external factors affecting corrosion, such as the mechanical and electrochemical properties of materials, stress distribution at the joint, and the flow condition in the pipe, and their combined action were investigated. Failure attribution was performed on this basis. Specifically, a hardness test was conducted on different zones of the joint, including the base metal (BM), heat-affected zone (HAZ) and weld metal (WM), to investigate the abrasive wear resistance at the joint. Microstructures of the three zones were characterized by optical microscopy. The morphology and composition of the corrosion film covering the inner surface of the welded joint were characterized by scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS) and X-ray powder diffraction (XRD). Potentiodynamic polarization experiments and electrochemical impedance spectroscopy (EIS) were conducted on samples from the three zones to investigate their electrochemical performances. Moreover, a computational fluid dynamics (CFD) analysis was performed to obtain the flow condition near the joint, and a finite element method (FEM) was used to calculate the stress distribution at the joint. The results showed that the internal corrosion mechanism of the pipeline was CO2 corrosion accelerated by detrimental Cl-. The geometric discontinuity of the welded joint was the main cause of the accelerated corrosion damage at the joint. The change in wall thickness of the welded pipe not only produced gas vortices but also led to stress concentration. Both of these conditions can further accelerate corrosion by destroying the protectiveness of the corrosion film. It is suggested that the corrosion can be prevented by eliminating the inhomogeneities of fluid flow and stress distribution. Keywords: Pipeline failures; Welds; Corrosion; 1. Introduction Welding is the most commonly used technology to connect steel natural gas pipelines. By the end of 2013, the total length of natural gas pipeline in China had reached 62000 km. With the extensive use of natural gas pipelines in China, corrosion failures at welded joints of natural gas pipes have begun to more frequently occur, causing substantial annual economic losses to enterprises and the country. Due to metal melting and solidification involved in the welding process, along with the melting and recrystallization of crystals, the material properties of a welded joint, which includes different zones, such as the weld metal (WM), heat-affected zone (HAZ) and base metal (BM), always exhibit inhomogeneities. Moreover, certain joint forms, such as T-joints and unequal wall thickness joints, always lead to geometric discontinuities, and this type of discontinuity can cause stress concentration at the joint and a sudden disturbance of fluid flow in the pipe. The inhomogeneities of material properties, stress  Corresponding author. Tel.: +86 02982665602. E-mail address: [email protected] (G. Cheng). 1

ACCEPTED MANUSCRIPT

D

MA

NU

SC R

IP

T

distribution discontinuity, and flow disturbance are considered to be a type of generalized “inhomogeneity” in this paper. Inhomogeneous distributions of matter and energy are manifested in different forms, and the inhomogeneity always becomes the driving force behind other natural phenomena. In regard to pipe welded joints, the inhomogeneities of material properties, stress and flow make the joints become the sensitive parts in terms of the internal corrosion of pipelines. For decades, increasingly more research has focused on corrosion damage at welded joints. L. M. Byvoishchik et al. [1] investigated the influence of the microstructure on the plasticity, impact toughness, and corrosion resistance of welded joints in low-alloy steels, which are widely used in the oil industry. W. Y. Wu et al. [2] studied the microstructure, mechanical properties, and corrosion behavior of dissimilar metal welded joints between carbon steel and ferritic stainless steel. The influence of the fluid flow on the preferential corrosion of welded X65 pipeline steel in brines containing carbon dioxide was studied by K. Alawadhi et al. [3,4] and M. A. Adegbite et al. [5], respectively, and further consideration was given to the influence of sand on preferential weld corrosion in such environments by R. Barker et al. [6]. Moreover, corrosion failures caused by stress, such as stress corrosion cracking, have also been a key concern of many scholars [7,8]. However, for welded joints in natural gas gathering pipelines, relatively little comprehensive analysis considering multiple factors affecting corrosion, such as the material properties, stress and flow, has been conducted. There has also been little research on their combined action. This paper studies a corrosion failure case of a welded joint in a certain natural gas gathering pipeline via experimental analysis and numerical calculation. Multiple sensitive factors affecting corrosion at the joint and their combined action were analyzed. The influence of inhomogeneity on the weld corrosion was discussed. On this basis, the failure mechanisms of weld corrosion were investigated. Two anti-corrosion measures were also provided to prevent similar corrosion failures.

TE

2. Natural gas gathering process and corrosion failure phenomena

AC

CE P

In the production of natural gas, gathering pipelines gather gases from different wells and transport them to a purification plant, where dehydration and desulfurization are conducted. Due to the direct contact between gathering pipelines and untreated gases that are highly corrosive, severe pipeline internal corrosion always occurs. Meanwhile, the flammability and high pressure of natural gas can greatly increase losses caused by corrosion failures. Serious corrosion damage at a welded joint of a gathering pipeline was detected in a gas field in Northeast China during routine maintenance. Fig. 1 shows the location of the failed welded joint in the pipe network. An image of the pipe section with the failed welded joint is shown in Fig. 2. It can be observed that internal corrosion occurred along the entire pipe section, with brownish-yellow corrosion products covering the pipe inner surface. However, compared with other parts of the pipe, the welded joint suffered more serious corrosion. The wall thinning at some locations along the circumference of the girth weld was significant, and perforations were about to occur. To ensure the safe production of natural gas, this dangerous pipe section was immediately replaced. This study attempts to determine what type of corrosion occurred on the inner surface of the gathering pipe, why the corrosion damage at the joint was more serious, and what the corrosion failure mechanism was at the joint. Through a field investigation, information about the pipe material, operating parameters, and components of the transported medium were obtained. The corrosion environment in the gathering pipe can be assessed accordingly. Similar to many pipelines in China, this failed pipe was made of 20G steel, which has a low carbon content and exhibits good ductility and toughness. Table 1 shows the standard chemical composition of the steel [9]. The design and operating parameters of the failed pipe are given in Table 2. Table 2 shows that there were some water and sands in the untreated natural gas. Table 3 and Table 4 show the chemical composition of the gas and water in the pipe, respectively. In addition, the field investigation also found that the failed joint was formed by butt welding two pipe sections with different specifications (Φ 60 mm × 9 mm and Φ 60 mm × 7 mm). The outer diameters of the two pipe sections were identical, but the wall thicknesses were different. It was ensured during welding that the outer 2

ACCEPTED MANUSCRIPT surfaces of the two pipe sections were flush with each other. Therefore, a mismatch of the inner surfaces appeared, and a small “step” was formed due to the different wall thicknesses. The transported medium flowed from the thick wall pipe to the thin wall pipe, and the inner diameter of the flow channel became larger after the fluid flowed past the welded joint.

T

3. Methods of failure mechanism analysis

AC

CE P

TE

D

MA

NU

SC R

IP

To identify the internal corrosion mechanism, a further assessment of the corrosion environment and a deep analysis of the corrosion reaction products must be performed. The cause of the more serious corrosion damage at the welded joint might be local material deterioration or local environmental degradation at the joint. Thus, material properties at the joint, such as electrochemical corrosion resistance and abrasive wear resistance, should be studied to determine whether local material deterioration has occurred. The factors that might cause local environmental degradation, such as flow conditions and stress distribution at the joint, should also be investigated. Therefore, the following methods were adopted to investigate the current corrosion failure case. The as-received failed pipe section was longitudinally cut in half, and a visual examination of the failed welded joint was conducted to assess the damage status. A microhardness test was conducted using a Vickers Hardness Tester to investigate the abrasive wear resistance against sand particles at different zones of the joint. A strip sample with different zones of the welded joint, including the BM, HAZ and WM, was cut out from the joint. The longitudinal cross section surface of the sample was hardness tested with a 1.961-N load and 15-s dwell time. There were three parallel hardness test lines on the sample surface along the axial direction of the pipe and across the BM, HAZ, and WM of the joint. The line spacing was 1.65 mm, and each line had 56 test points with a point interval of 0.5 mm. The metallographic examination was performed on different zones of the welded joint using optical microscopy to obtain microstructures corresponding to the properties of different zones. The linear intercept method was used to estimate the grain sizes of different microstructures. Corroded samples cut from the failed welded joint were analyzed by scanning electron microscopy (SEM). The morphology of the corroded surface was examined to investigate the quality and adhesion of the corrosion product film on the surface of the joint and to then assess the film protectiveness accordingly. Energy-dispersive spectroscopy (EDS) analysis was used to identify the elemental chemical composition of the corrosion product film. The film was then removed using hydrochloric acid. The sample without film was reexamined using SEM to observe the morphology of the corroded metal substrate. To qualitatively and quantitatively identify the compounds in the corrosion products, the products from the severely corroded area at the joint and the products from the slightly corroded area away from the joint were both analyzed using X-ray diffraction (XRD). The diffraction angle ranged from 0.5° to 120° with an angle step of 0.02°. Electrochemical tests of materials from different zones of the welded joint were conducted to investigate their electrochemical behaviors and corrosion resistances in the simulated corrosion environment in the pipe. All zones of a strip sample from the failed welded joint, including the base metal of the thick wall pipe (BM_A), heat-affected zone of the thick wall pipe (HAZ_A), weld metal (WM), heat-affected zone of the thin wall pipe (HAZ_B), and base metal of the thin wall pipe (BM_B), were distinguished and marked, as shown in Fig. 3. Independent metal samples from each zone were cut out and fabricated into working electrodes for the electrochemical tests. The electrolyte solution used for the electrochemical tests was prepared with NaHCO3, NaCl, Na2SO4 (all analytical reagent grade) and deionized water according to Table 4. Prior to each test, the electrolyte solution was purged with a mixed gas of N2 and CO2 for 24 hours to eliminate the influence of O2 and to adjust the solution pH to approximately 7.3, namely, the pH of the water in the gathering pipeline, as listed in Table 4 [10]. All tests were performed at a pressure of 0.1 MPa. The test temperature was maintained at 46 ℃, the operating temperature of the gathering pipe. A three-electrode cell was used to perform electrochemical tests with a platinum 3

ACCEPTED MANUSCRIPT

MA

NU

SC R

IP

T

plate as the counter electrode, a saturated calomel electrode (SCE) as the reference electrode, and samples prepared above as working electrodes. After the open-circuit potential of each working electrode was stable, the electrochemical impedance spectrum (EIS) measurement and the potentiodynamic polarization experiment were performed according to ISO/TR 16208:2014 [11] and ISO 17475:2005 [12], respectively. During the EIS measurement, a 10-mV sinusoidal signal was applied to the electrode at the open-circuit potential, and the measuring frequency ranged from 105 Hz to 0.01 Hz. In the potentiodynamic polarization experiment, the potential was scanned from -250 mV to +250 mV versus an open-circuit potential at a scan rate of 0.5 mV/s. The natural gas flow in the pipe section with the unequal wall thickness welded joint was numerically simulated using computational fluid dynamics (CFD) methods. The influence of the sudden change in wall thickness on the flow characteristics was investigated. Because the geometric details of the uncorroded and newly welded joint that was just put into service cannot be acquired, the flow channel was simplified to a 2D axisymmetric model in which the welded joint was replaced by a diverging pipe section. The computational domain and mesh are shown in Fig. 4. Inflation layers were used to model the flow within the boundary layer. Model settings of the simulation are listed in Table 5. It is worth mentioning that the inlet in the thick wall pipe section was set away from the welded joint to ensure a fully developed flow. A steady-state solution was obtained after the calculation. The stress distribution at the welded joint was analyzed using finite element methods (FEM). The influence of the stress concentration on corrosion due to the geometric discontinuity of the joint was investigated. A 3D linear elastic model of the pipe with the welded joint was established, as illustrated in Fig. 5. The pipe inner surface was subjected to an internal pressure, p , of 18.2 MPa. A tensile stress of magnitude pDi,2 B  Do,2 B  Di,2 B  was

D

applied on the end face of the thin wall pipe. Di, B and Do, B denote the inner and outer diameters of the thin wall

4. Results

CE P

TE

pipe, respectively. This tensile stress represents the axial force along the pipe due to the end cap effect. A zero axial displacement was imposed on the end face of the thick wall pipe. The welded joint was assumed to have the same Young’s modulus (209 GPa) and Poisson’s ratio (0.28) as the 20G pipes. The model consisted of 3,106,717 elements, with local mesh refinement at the joint. After numerical calculation, the stress distribution was obtained.

AC

4.1. Multiphase corrosion environment in the pipeline As mentioned earlier, in addition to the natural gas medium, water and sands were also observed in the gathering pipeline. Although no specific sand content is given in Table 2, the field investigation showed that a solid particle phase, such as sands, was indeed found in the pipe. These solid particles can impact the pipe inner surface and result in a more severe pipe environment. Thus, a multiphase corrosion environment including a gas phase, liquid phase, and solid phase formed in the gathering pipeline. Table 3 shows that N2, CO2, and trace He were found in the gas medium in addition to natural gas. Among these gases, CO2, a typical corrosive medium in oil and gas pipes, can be dissolved in water and result in a corrosive solution environment. Table 4 shows that the water solution in the gathering pipe was nearly neutral, with substantial amounts of HCO3- and Cl- and a small amount of SO42- in it. 4.2. Damage status of the failed welded joint assessed by visual examination The damage status of the failed welded joint is shown in Fig. 6. It can be clearly observed that the welded joint was an unequal wall thickness joint. In the half pipe shown in Fig. 6, there was a corrosion groove at the failed joint, with its widest width of approximately 8.78 mm and its deepest point 5.96 mm lower than the inner surface of the thin wall pipe. On the cutting plane of the pipe, the remaining wall thicknesses at the joint were 7.24 mm and 4.42 mm, respectively. Therefore, the corrosion at the welded joint was extremely serious. It could have lead to a catastrophic accident if shutdown and replacement were not promptly performed. At the locations away 4

ACCEPTED MANUSCRIPT from the welded joint in the pipe (BM area), uniform wall thinning occurred, and the inner surface was covered with brownish-yellow corrosion products. Characteristics of uniform corrosion were shown at these locations. 4.3. Hardness measurements

IP

T

The micro-Vickers hardness distribution at the welded joint is shown in Fig. 7. The average hardness value of the WM was approximately 197 HV, and the hardness value of the BM ranged from 130 HV to 150 HV. It can be observed that the WM had the highest hardness value and that there was a significant increase in hardness from the BM to the WM. The hardness value of the HAZ was between that of the BM and that of the WM.

SC R

4.4. Microstructure inhomogeneity analysis

D

MA

NU

Optical microscopy of the welded joint consisting of the BM, HAZ, and WM is depicted in Fig. 8. The microstructure of the BM, mainly composed of white ferrite and black lamellar pearlite, was a typical metallographic structure of hypoeutectoid steel. The average grain size of the BM was approximately 15 μm. In the HAZ, the grains became noticeably larger at the sites near the fusion line. The average grain size of the HAZ was approximately 111 μm. The proeutectoid ferrite precipitated along the prior coarse austenite grain boundaries. A mixture of pearlite and acicular ferrite appeared in the grain interior. Obvious features of Widmanstatten structure were shown. The grain structure of the WM consisted of columnar grains with an average transverse grain size of approximately 122 μm and an average longitudinal size larger than 420 μm. As indicated in Fig. 8(c), coarse acicular and massive proeutectoid ferrite distributed at the boundaries of coarse massive grains. Moreover, acicular ferrite grew into grain interiors. Therefore, the microstructure of the grain interior of the WM was composed of pearlite and acicular ferrite.

TE

4.5. SEM, EDS, and XRD analysis of the failed welded joint

AC

CE P

The SEM morphology of the corrosion product film on the inner surface of the welded joint where the pipe wall thickness changes is presented in Fig. 9 and Fig. 10. As shown in Fig. 9, a corrosion product film with a loose and porous structure was found at the location where accelerated damage occurred. From Fig. 10(b), it can be observed in a longitudinal sectional view that the corrosion product film showed clearly different characteristics at different locations. Accordingly, the SEM view in Fig. 10(b) can be divided into three areas: slightly corroded area, severely corroded area, and the transition area from slightly corroded area to severely corroded area. As for the film quality, the film in slightly corroded area was thicker and more compact, with nearly no pores and cracks under the current magnification. In the transition area, pores of various sizes emerged, indicating the decrease of film quality. In the severely corroded area, a substantial decrease in the film thickness was observed on the surface of the diverging pipe section. Cracks and pores were also distributed in the film. Of all the three areas, the severely corroded area exhibited the worst film quality. Another important property to evaluate the protectiveness of the film is the film adhesion. Clearly, the film in the severely corroded area had the poorest adhesion. At the most severely corroded locations in Fig. 10(b), the corrosion film was entirely detached from the steel substrate, indicating that the steel substrate was completely exposed to the erosive corrosive environment. In contrast, the film in the slightly corroded area showed good adhesion. Fig. 11 compares the protectiveness of the corrosion film from different areas under a higher magnification. Similarly, the thick, compact and well-adhered film in the slightly corroded area indicated that it had the best protectiveness. The film in the severely corroded area was thin, porous, badly-adhered, and sometimes discontinuous. Thus, it showed the worst protectiveness. The protectiveness of the film in the transition area was in the middle. The morphology of the corroded metal substrate after the removal of the surface film is shown in Fig. 12. From Fig. 10(b), Fig 11, and Fig. 12, it can be observed that neither intergranular nor transgranular micro cracks appeared in the metal substrate, indicating that stress corrosion cracking (SCC) did not occur. Fig. 13 shows the results of the EDS analysis of the corrosion product film. Apparently, the main elements of 5

ACCEPTED MANUSCRIPT

IP

T

the corrosion products were Fe, O, C, Si, Al, Ca, and K. Small amounts of Mg, Mn and S were also found in the film. Among these elements, Fe, C, Si, and Mn were the elements contained in 20G steel. Elements such as Al, K, Ca, Mg, and S might originate from the residual medium in the pipe or other surface residual material that had not been cleaned up. The results of the XRD analysis are shown in Fig. 14. The products from both areas had the same principal constituents: FeCO3, Fe2O3, and FeO(OH). At the joint, the contents of these compounds were 56.07 %, 39.91 % and 4.02 % in mass fraction, respectively. At the locations away from the joint, their contents were 79.36 %, 12.61 % and 8.03 % in mass fraction, respectively.

SC R

4.6. Electrochemical tests

D

MA

NU

Obtained by potentiodynamic measurements, the polarization curves of samples from different zones of the welded joint are depicted in Fig. 15. Table 6 shows the electrochemical kinetic parameters obtained by Tafel extrapolation for all samples. Due to the non-linearity of the anodic curves, only the cathodic curves were used in the calculation. It can be observed from Fig. 15 that all the polarization curves were highly consistent with each other. The differences among the curves were quite small. Similarly, Table 6 shows that the corrosion current densities (i.e., corrosion rates) of different samples were very close to each other. All the corrosion current densities were very low and no obvious increase in corrosion current density was found at the joint. Fig. 16 shows the distribution of the corrosion potential at the welded joint. No apparent decrease in the corrosion potential occurred at the failed joint. Fig. 17 shows the impedance diagrams of samples from different zones of the welded joint. It can be seen that all samples had similar polarization resistances. This means that no obvious decrease in corrosion resistance occurred at the failed joint.

TE

4.7. CFD and FEM analysis

AC

CE P

The natural gas streamlines near the welded joint in the pipe are shown in Fig. 18(a). The streamline color represents the gas velocity magnitude. As shown in Fig. 18(a), the average gas velocity in the thin wall pipe section was slightly lower than that in the thick wall pipe section. This was due to an increase in the pipe inner diameter and a constant mass flow rate of the gas. More importantly, at the downstream end of the diverging pipe section, which represents the unequal wall thickness joint, a back flow region formed, and a gas vortex was produced. Fig. 18(b) shows the velocity vectors of the gas vortex. It can be observed that the gas flow velocity here significantly decreased. It should be noted that in the current pipe, the liquid phase had a very low content of approximately 0.18 % in volume fraction compared with gas phase. Therefore, the liquid in the pipe was entrained in the gas flow in the form of liquid droplets. Similarly, the solid particles were also entrained in the gas flow and moved with the gas. Fig. 19 shows the Von Mises stress distribution at the welded joint. In general, the stress in the pipe wall decreased radially outward. The average stress in the thick pipe wall was lower than that in the thin pipe wall. Due to the internal pressure, obvious stress concentration was found on the inner surface at the inlet end of the thin wall pipe section, where the maximum Von Mises stress of approximately 129.18 MPa was developed. This value was approximately 3.33 times higher than the Von Mises stress at the mid-thickness of the thick wall pipe and approximately 2.24 times higher than that of the thin wall pipe. 5. Discussions 5.1. Internal corrosion mechanisms of the natural gas gathering pipeline From the analysis in section 4.1.1, a multiphase medium including gas, liquid and solid existed in the gathering pipeline. CO2 and HCO3- were found in the gas phase and liquid phase, respectively. The results of the EDS analysis show that the main elements in the corrosion product film included Fe, C, and O. A further XRD analysis indicates that FeCO3 was the main compound in the corrosion product. All of this information indicates 6

ACCEPTED MANUSCRIPT that the gathering pipe was suffering from CO2 corrosion, a common type of corrosion in oil and gas pipelines. When CO2 is dissolved in water, the following reactions occur in the solution: (1) CO2 + H2OH2CO3 H2 CO3  H+ + HCO3-

(2)

HCO  H + CO3

(3)

3

+

2-

IP

T

H+, H2CO3, and HCO3- can act as the depolarizers of electrochemical corrosion reactions, and the following cathodic reactions occur. (4) 2H+ + 2e  H2 2HCO + 2e  H2 + 2CO3 3

SC R

2H 2 CO3 + 2e  H2 + 2HCO32-

(5) (6)

MA

NU

Among these reactions, reaction (6) becomes important when the solution pH>5 [13]. The water pH in the gathering pipe was approximately 7.3. Thus, the direct reduction of HCO3- is also a main cathodic reaction. The dominant anodic reaction is the dissolution of iron in the steel: (7) Fe  Fe2+ + 2e 2+ 2In addition, Fe and CO3 ions in the solution can combine to form solid FeCO3 film covering the steel surface: (8) Fe2+ + CO32-  FeCO3

AC

CE P

TE

D

Fe2O3 and FeO(OH), the other two compounds in the surface film determined by XRD analysis, are typical products of the atmospheric corrosion of steel [14]. However, it is noted from Table 3 that there was no oxygen in the gathering pipe. Hence, theoretically, these two compounds could not be produced. However, the field investigation found that before the failed pipe section was sent for inspection, it had not been kept properly sealed by the gas field staff and had been exposed to the atmosphere for a short time. Considering that the contents of FeO(OH) and Fe2O3 were small, these two compounds are deduced to be produced due to improper preservation before the tests, and they are not regarded as the internal corrosion products of the gathering pipeline. In addition to CO2 and HCO3-, the water in the pipe also contained substantial amounts of Cl-. Various studies have shown that a certain concentration of Cl- can promote CO2 corrosion [15]. This is because (1) Cl- can promote the dissolution of iron through a catalytic mechanism [16] and because (2) Cl- can aggregate on the interface of the steel substrate and surface film [17], and reduce the protectiveness of the film [15]. Therefore, Cl- can also contribute to the damage on the entire internal surface of the pipe. In conclusion, it can be deduced that the internal corrosion of the natural gas gathering pipeline was CO2 corrosion accelerated by detrimental Cl-. 5.2. The accelerated corrosion mechanism of the inhomogeneous welded joint 5.2.1. Inhomogeneities of material properties and their effects As shown in the hardness tests results, the material hardness gradually increased from the BM to the HAZ to the WM. This means that as for the metal substrate of the current pipe, materials from the WM and the HAZ have a better capacity to endure mechanical damage from solid particles compared to material from the BM. As a result, it can be excluded that the pure abrasive wear from solid particles caused more serious damage at the welded joint than at other parts of the pipe. However, due to the electrochemical corrosion, what directly contacts with solid particles was the corrosion product film rather than the metal substrate. Thus, the advantage that the material from the welded joint has a higher abrasive wear resistance due to its higher hardness cannot be reflected. Under this circumstance, the film protectiveness plays a more important role. From an engineering view point, the samples from different zones of the failed joint did not present remarkable inhomogeneities in electrochemical properties. Electrochemical tests results show that in the simulated water solution, great increase in corrosion current density did not occur at the WM and the HAZ. An obvious decrease in electrochemical corrosion resistance was not found at the joint, and thus did not cause the accelerated 7

ACCEPTED MANUSCRIPT

SC R

IP

T

weld corrosion. In addition, there was no obvious decrease in corrosion potential at the failed joint. The maximum difference among the measured corrosion potentials at the joint was even less than 18 mV. Thus, the galvanic effect among different zones at the joint can be ignored. Generally, the galvanic effect is considered to be significant when the corrosion potential difference in a galvanic couple equals or exceeds 100 mV ~ 130 mV [18]. Moreover, corrosive media, such as CO2 and Cl-, have to be dissolved into water before they react with the pipe steel. As mentioned above, the liquid in the gas pipe existed in the form of droplets. That means that the electrochemical corrosion was mainly caused by corrosive liquid droplets. Therefore, there was not a very good ionic conductivity to cause galvanic corrosion. Hence, the probability of the occurrence of galvanic corrosion was very low. In general, the inhomogeneities of material properties did not greatly affect the damage at the joint.

CE P

TE

D

MA

NU

5.2.2. Inhomogeneity of fluid flow and its effect The special geometry of the failed joint resulted in a diverging pipe section. One of the most prominent influences of this geometric discontinuity was the formation of a gas vortex at the joint, as shown in the CFD analysis. As mentioned above, the contact of the corrosive liquid droplets and the steel can cause electrochemical corrosion. Moreover, the friction and collisions of the droplets and solid particles in the gas flow with the pipe wall can result in mechanical damage on this basis. The appearance of a gas vortex means that the droplets and particles entrained in the gas would also circle round at the joint with the gas. This can no doubt intensify the contacts and collisions between the steel at the joint and the droplets or particles in the pipe. Electrochemical corrosion products can be produced when the corrosive droplets contact with the steel at the joint. However, due to the collisions of the particles and droplets, the products can not form a complete and compact film. From the SEM analysis, it can be found that the severely corroded area with a poor-quality, detached and non-protective film was right beside the gas vortex. Therefore, it is reasonable to regard that the erosion corrosion from the droplets and particles in the gas vortex was one of the main causes of the bad film condition at the joint. The film thinning, the pores and cracks in the film, and the film detachment, can all result from the erosion corrosion due to the gas vortex. In addition, the decrease in gas velocity at the vortex position can increase the residence time of corrosive droplets and solid particles at the joint and make their interactions with the pipe steel more sufficient; this would then aggravate the damage at the joint.

AC

5.2.3. Inhomogeneity of stress distribution and its effect The geometric inhomogeneity at the welded joint can also lead to stress concentration at the joint. The results of the FEM analysis indicate that stress concentration occurred at the inlet end of the thin wall pipe section, where the inner diameter of the pipe no longer increases. It is also the exact location of the severely corroded area with a poor-quality, detached and non-protective film, as shown in the SEM analysis. It is worth noting that although only the operating stress was considered in the current analysis, relatively large welding residual stress in the early service stage of the pipe can also produce negative effects on the joint. Stress can affect corrosion by changing the properties of the corrosion product film. Subject to the gas pressure in the pipe, both the steel substrate and the corrosion product film on it deform. However, the deformation of the “brittle” product film of CO2 corrosion is incompatible with that of the “plastic” substrate metal [19]. This is particularly true at those stress concentration locations where the resulting high strain can aggravate the damage to the corrosion product film and decrease its protectiveness. Therefore, numerous pores and cracks appeared in the corrosion film in the severely corroded area at the joint. In some places, the film was interupted, or even detached from the steel substrate, as illustrated in Fig. 10 and Fig. 11. The film at the stress concentration locations exhibited poor protectiveness. 5.2.4. Combined action of fluid flow and stress distribution As mentioned above, the inhomogeneities of material properties due to the welding process did not greatly affect the damage at the joint. However, the inhomogeneities of fluid flow and stress distribution due to the special joint geometry had a significant influence. 8

ACCEPTED MANUSCRIPT

MA

NU

SC R

IP

T

When the metal corrosion is accompanied by the formation of a protective corrosion product film, the corrosion rate depends on the following stages: (1) the transport of the reactant to the metal surface through corrosion product film; (2) the corrosion reaction; (3) the release of the corrosion products from the metal surface. At the failed joint under study, fluid flow and stress jointly acted on the corrosion product film. Through collisions of droplets and particles, and the action of the stresses within the film, the film integrity and adhesion were destroyed. The film in some severely corroded area was even entirely detached and provided nearly no protectiveness. Therefore, at the joint, corrosive reactants in the droplets can easily reach the steel surface, with little or even no barrier. Thus, the first stage of the corrosion process can be greatly accelerated. Moreover, the exposed metal substrate at the joint means that, as soon as new products were formed at the joint, they were taken away from the surface. There was no opportunity for them to form a complete film. Therefore, the release of the corrosion products (i.e., the third stage in the corrosion process) was also accelerated. In all, poor film protectiveness at the joint accelerated the mass transfer involved in corrosion, and eventually accelerated the corrosion rate. Without the protection from the film, the exposed metal substrate can also suffer from the erosion of solid particles. Fig. 20 shows the damage mechanism of the unequal wall thickness welded joint under study. Special joint geometry led to a harmful gas vortex and stress concentration at the joint, which in turn resulted in bad film protectiveness. Long term exposure of the metal substrate to the erosive corrosive environment at the joint promoted the damage, and eventually made the joint a vulnerable part in the pipe. 5.3. Prevention approaches

6. Conclusions

CE P

TE

D

Elimination of the inhomogeneities of fluid flow and stress distribution should be conducted to prevent the accelerated corrosion at the joint. Details are as follows: (1) An appropriate joint form should be designed. The use of unequal wall thickness welded joints should be avoided to eliminate the geometric inhomogeneity and to ensure good stress distribution and flow conditions at the welded joint. (2) A post-weld heat treatment should be applied to reduce the welding residual stress at the joint.

AC

Serious corrosion failure occurred at a welded joint of a natural gas gathering pipeline in Northeast China. The unequal wall thickness welded joint suffered substantially more severe corrosion damage than did other parts of the pipe. To determine the failure causes, a comprehensive analysis was conducted based on the inhomogeneities of material properties, fluid flow, and stress distribution. The main research conclusions are as follows: (1) The internal corrosion of the failed gathering pipe was CO2 corrosion accelerated by detrimental Cl-. (2) The inhomogeneities of material properties due to the welding process did not greatly affect the damage at the joint. (3) The special joint geometry resulted in a gas vortex and led to stress concentration. Due to the erosion corrosion of the droplets and particles in the vortex, and the film internal stress caused by stress concentration, the protectiveness of the corrosion film was destroyed. (4) Poor film protectiveness can promote corrosion by accelerating mass transfer process. Long term exposure of the steel substrate to erosive corrosive environment made the joint vulnerable. (5) Inhomogeneities of fluid flow and stress distribution due to the geometric discontinuity were the main reasons of the corrosion failure. To prevent the damage, an appropriate joint form should be used. A post-weld heat treatment is also helpful.

9

ACCEPTED MANUSCRIPT Acknowledgments This project was supported by the National Natural Science Foundation of China (Grant No. 51175403) and PetroChina Innovation Foundation (No. 2014D-5006-0606).

T

References

IP

[1] Byvoishchik LM, Luchkin RS, Platonov SY. Structural factor of the corrosion and mechanical strength of welded joints in oil transmission pipes. Weld Int 2010;24:23–8.

SC R

[2] Wu WY, Hu SS, Shen JQ. Microstructure, mechanical properties and corrosion behavior of laser welded dissimilar joints between ferritic stainless steel and carbon steel. Mater Design 2015;65:855–61.

[3] Alawadhi K, Robinson MJ. Preferential weld corrosion of X65 pipeline steel in flowing brines containing carbon dioxide. Corros Eng Sci Technol 2011;46:318–29.

[4] Alawadhi K, Aloraier AS, Joshi S, Alsarraf J, Swilem S. Investigation on preferential corrosion of welded carbon steel under

NU

flowing conditions by EIS. J Mater Eng Perform 2013;22:2403–10.

[5] Adegbite MA, Robinson MJ, Impey SA. The influence of hydrodynamics on the preferential weld corrosion of X65 linepipe steel in flowing brine containing carbon dioxide. In: NACE International. Corrosion 2014 Proceedings; 2014.

MA

[6] Barker R, Hu XM, Neville A. The influence of high shear and sand impingement on preferential weld corrosion of carbon steel pipework in CO2-saturated environments. Tribol Int 2013;68:17–25. [7] Saleem B, Ahmed F, Rafiq MA, Ajmal M, Ali L. Stress corrosion failure of an X52 grade gas pipeline. Eng Fail Anal 2014;46:157–65.

D

[8] Chauhan V, Fordyce I, Gilliver J, Peravali S, Connell A, Thompson I, et al. Failure investigation of a natural gas transmission ASME; 2012. p. 637–51.

TE

pipeline. In: Proceedings of the 2012 9th International Pipeline Conference, vol. 2; 2012 Sep 24–28; Calgary, Canada. New York: [9] GB 5310–2008. Seamless steel tubes and pipes for high pressure boiler. Beijing; 2008.

CE P

[10] Zhang GA, Cheng YF. Micro-electrochemical characterization of corrosion of welded X70 pipeline steel in near-neutral pH solution. Corros Sci 2009;51:1714–24.

[11] ISO/TR 16208:2014. Corrosion of metals and alloys–Test method for corrosion of materials by electrochemical impedance measurements. 2014.

[12] ISO 17475:2005. Corrosion of metals and alloys–Electrochemical test methods–Guidelines for conducting potentiostatic and

AC

potentiodynamic polarization measurements. 2005. [13] Nordsveen M, Nesic S, Nyborg R, Stangeland A. A mechanistic model for carbon dioxide corrosion of mild steel in the presence of protective iron carbonate films–part1: theory and verification. Corrosion 2003;59:443–56. [14] Han SC. Atlas of microstructure on metals corrosion. Beijing: National Defense Industry Press; 2008. [15] Liu QY, Mao LJ, Zhou SW. Effects of chloride content on CO2 corrosion of carbon steel in simulated oil and gas well environments. Corros Sci 2014;84:165–71. [16] Burstein GT, Davies DH. The effects of anions on the behavior of scratched iron electrodes in aqueous solutions. Corros Sci 1980;20:1143–55. [17] Mccafferty E. A competitive adsorption model for the inhibition of crevice corrosion and pitting. J Electrochem Soc 1990;137:3731–7. [18] Huerta EO. Corrosion and degradation of materials. Madrid: Síntesis; 1997. [19] Ramachandran S, Campbell S, Ward MB. The interactions and properties of corrosion inhibitors with byproduct layers. In: NACE International. Corrosion 2000 Proceedings; 2000.

10

ACCEPTED MANUSCRIPT Figure captions Fig. 1. Location of the failed welded joint in the pipe network. Fig. 2. Pipe section with the failed welded joint (a longitudinal sectional view).

T

Fig. 3. Different zones of the welded joint (A: thick wall pipe; B: thin wall pipe).

IP

Fig. 4. Computational domain and mesh of the CFD calculation.

Fig. 6. Damage status of the failed welded joint (in mm). Fig. 7. Micro-Vickers hardness distribution at the welded joint.

SC R

Fig. 5. FEM model of the pipe: (a) loads and constraints; (b) mesh model (a longitudinal sectional view).

Fig. 8. Microstructures at the welded joint (200×): (a) BM; (b) HAZ; (c) WM.

NU

Fig. 9. SEM morphology of the corrosion product film on the inner surface of the failed joint (a top view). Fig. 10. SEM morphology of the corrosion product film on the inner surface of the failed joint (a longitudinal

MA

sectional view): (a) location of the SEM analysis; (b) SEM morphology of the corrosion film. Fig. 11. Comparison of the protectiveness of the corrosion film: (a) slightly corroded area; (b) transition area from slightly corroded area to severely corroded area; (c) severely corroded area.

D

Fig. 12. SEM morphology of the corroded steel substrate after the removal of the surface film.

TE

Fig. 13. EDS analysis of the corrosion product film: (a) location of the EDS analysis; (b) EDS spectrum of the corrosion product film.

Fig. 14. XRD analysis of corrosion products: (a) XRD spectrum of corrosion products from the severely corroded

CE P

area at the joint; (b) XRD spectrum of corrosion products from the slightly corroded area away from the joint. Fig. 15. Polarization curves of samples from different zones of the joint. Fig. 16. Corrosion potential distribution at the welded joint.

AC

Fig. 17. EIS spectra of samples from different zones of the welded joint. Fig. 18. Flow condition in the pipe: (a) natural gas streamlines near the welded joint; (b) velocity vectors of the gas vortex.

Fig. 19. Stress distribution at the welded joint (a longitudinal sectional view). Fig. 20. Damage mechanism of the unequal wall thickness welded joint.

11

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 1

12

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 2

13

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

Fig. 3

14

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

Fig. 4

15

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

Fig. 5

16

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

Fig. 6

17

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 7

18

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 8a

19

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 8b

20

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 8c

21

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 9

22

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 10a

23

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 10b

24

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 11a

25

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 11b

26

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 11c

27

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 12

28

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 13a

29

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 13b

30

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 14

31

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 15

32

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 16

33

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 17

34

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

Fig. 18a

35

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

Fig. 18b

36

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 19a

37

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

Fig. 19b

38

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Fig. 20

39

ACCEPTED MANUSCRIPT Table 1 Standard chemical composition (wt.%) of 20G steel. C

Si

Mn

P

S

Fe

Content

0.17~0.23

0.17~0.37

0.35~0.65

≤0.025

≤0.015

Bal.

AC

CE P

TE

D

MA

NU

SC R

IP

T

Element

40

ACCEPTED MANUSCRIPT Table 2 Design and operating parameters of the failed pipe. Pipeline

Operational

Design

Operating

Design

Operating

Daily gas

Daily water

pressure

pressure

temperature

temperature

production

production

(MPa)

(MPa)

(℃)

(℃)

(m3)

(m3)

29

18.2

45

46

Pipe specification

Material

date

Sand

medium (mm)

(yyyy-mm-dd)

Φ60×7

10.8×104

1.2

>0

CE P

TE

D

MA

NU

SC R

IP

gas

T

2007-08-25

AC

Φ60×9

Natural 20G

content

41

ACCEPTED MANUSCRIPT Table 3 Chemical composition (vol.%) of the gas in the gathering pipe. Methane

Ethane

Propane

Isobutane

N-butane

He

N2

CO2

Content

88.705

1.555

0.124

0.010

0.027

0.047

6.658

2.874

AC

CE P

TE

D

MA

NU

SC R

IP

T

Component

42

ACCEPTED MANUSCRIPT Table 4 Chemical composition of the water in the gathering pipe. Cl- content (mg/L)

SO42- content (mg/L)

pH

Water hardness (mg/L)

3616.85

2093.12

107.39

7.3

15.66

AC

CE P

TE

D

MA

NU

SC R

IP

T

HCO3- content (mg/L)

43

ACCEPTED MANUSCRIPT Table 5 Model settings of the gas flow simulation. Turbulence

Boundary conditions

model

Inlet

Outlet

Pipe inner surface

velocity inlet

pressure outlet

wall

5.56 m/s (calculated according

18.2 MPa (operating

non-slip boundary

to the data in Table 2)

pressure of the pipe)

standard wall function

momentum conservation

k-epsilon

axial

AC

CE P

TE

D

MA

NU

SC R

energy conservation

Symmetry axis

IP

mass balance

T

Governing equations

44

ACCEPTED MANUSCRIPT Table 6 Electrochemical kinetic parameters of samples from different zones of the welded joint. Zones Parameters

147.65 2

WM

146.68 -6

135.27 -6

BM_B

138.98

5.5463×10

-6

Corrosion current density icorr (A/cm )

3.7043×10

Corrosion potential Ecorr (mV vs SCE)

-779.34

-791.29

-796.62

-782.62

Corrosion rate CR (mm/y)

0.043

0.051

0.064

0.042

136.28

3.6104×10

-6

3.9536×10-6 -792.03 0.046

AC

CE P

TE

D

MA

NU

SC R

4.4245×10

HAZ_B

T

Cathodic Tafel slope bc (mV/dec)

HAZ_A

IP

BM_A

45

ACCEPTED MANUSCRIPT Highlights More severe corrosion occurred at an unequal wall thickness welded joint in a pipe.



The weld corrosion was studied based on inhomogeneity.



Material properties, fluid flow, and stress distribution at the joint were studied.



Geometric discontinuity was the main cause of the damage at the joint.



Flow and stress jointly promoted the damage by destroying film protectiveness.

AC

CE P

TE

D

MA

NU

SC R

IP

T



46