- Email: [email protected]
Failure of 321 stainless steel heater tube in heavy crude oil
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
Contents lists available at ScienceDirect
Case Studies in Engineering Failure Analysis journal homepage: www.elsevier.com/locate/csefa
Short communication
Failure of 321 stainless steel heater tube in heavy crude oil ⁎
MARK
H.M. Shalaby , K. Ravindranath, N. Tanoli, B. Al-Wakaa Petroleum Research Center, Kuwait Institute for Scientific Research, P.O. Box 24885, 13109-Safat, Kuwait
AR TI CLE I NF O
AB S T R A CT
Keywords: Stress corrosion cracking Charge heaters Refineries Sulphidation Salt deposition
Failure investigation was done on a 321 stainless steel charge heater tube which failed in a refinery unit processing heavy crude oil. Crude oil was the charge in the radiant and convection sections; while saturated stripping steam is present in convection section. After a leak was detected, visual inspection revealed that nine convection tubes had black oil/coke deposits on their external surfaces. The deposits were seen on the first three rows of tubes. When one of the tubes was lightly ground at the black colored area, a circumferential crack was visually observed. The investigation revealed that long-term aging, coupled with localized deposition of salts and coke from the heavy crude led to sensitization of the tube surface layers. This in turn resulted in sulphidation of the internal surface grain boundaries, formation of grooves, and cracking of the material. Thus, cracking was intergranular in nature in the initial stage, but became transgranular at later stages. It was concluded that cracking was due to chloride stress corrosion cracking catalyzed by the presence of sulphur-bearing species. It was recommended that the desalter operation be improved and frequent decoking and scale removal be carried out, with emphasis on the convection section at the refinery.
1. Introduction Stabilized austenitic stainless steels (SSs) have since been widely used in the petroleum refining industry because of their resistance to sulphidic corrosion and polythionic acid stress corrosion cracking (SCC), as well as of their excellent strength and toughness [1,2]. However, in some instances, stabilized austenitic SSs have been found to be susceptible to pitting, crevice corrosion, and SCC [3–5]. The susceptibility to SCC occurs under conditions involving high stress, changes in metallurgical structure due to high temperature exposure, and the presence of specific chemicals that promote cracking [6]. It was found that the factors most affecting corrosion of SSs in the refinery industry are chloride and hydrogen sulphide [3]. Hydrogen sulphide is one of the constituents of crude oils and refinery sour waters and is also formed by the decomposition of organic sulfur compounds at elevated temperatures [7]. Chloride SCC of austenitic SSs is transgranular, but can be intergranular if the material is sensitized [8]. There are reported failure cases suggesting that the presence of hydrogen sulphide (H2S) synergistically promotes chloride SCC [9,10]. Recent changes in refinery feedstocks in terms of increased H2S content have accelerated corrosion failures of stabilized austenitic SSs equipment. The present paper presents an example of such failures, where a 321 SS charge heater tube carrying heavy sour crude oil suffered cracking. 2. Background A leak occurred in a heater of a refinery unit handling heavy sour crude oil. Visual inspection revealed that eight convection tubes
⁎
Corresponding author. E-mail address: [email protected] (H.M. Shalaby).
http://dx.doi.org/10.1016/j.csefa.2017.04.004 Received 26 March 2017; Accepted 28 April 2017 Available online 20 May 2017 2213-2902/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 1. Camera photographs showing (a) as-received tube section; (b) circumferential crack seen after removal of studs; (c) deposited black layer on internal surface; and (d) stud portion of tube.
made of 321 SS had black oil/coke deposits on their external surfaces. The deposits were seen on the first two rows of plain tubes. The observations were made through available gaps between tubes, as access to the convection tubes was severely limited. In order to identify the leaky tubes, the refinery engineers pressurized the heater. However, no leak was detected as seen from the radiant zone. The eight plain tubes having black deposits were removed from the heater, which facilitated inspection of stud tubes. One stud tube in the third row was found to have black deposits. This tube was also removed, and the removed nine tubes were replaced with new ones. The removed 321 SS tubes were examined using dye penetrant test. No indication was found on all tubes. However, when the stud tube was lightly grounded externally at the black colored area, a circumferential crack was visually observed (Fig. 1a and b). The presence of this crack raised the possibility that the remaining heater tubes in operation may also have cracks. The heater is part of a heavy crude topping unit. The crude was the charge (tube side) in the radiant and convection sections since its commissioning over twenty years ago. Saturated stripping steam is present in the convection section. The temperatures at the inlet radiant and convection are 130 °C (266°F). On the other hand, the outlet temperatures are 345 °C (653°F) and 316 °C (601°F). The heavy crude charge has gravity in the order of 17 American Petroleum Institute (API), a total sulphur content of 4.48% wt, and organic chloride content of 3 mg/kg. However, this crude was changed to another heavy crude oil, a year earlier than the date of the current failure. This latter crude has 20.48 API gravity, a total sulphur content of 3.91% wt, and organic chloride content of 1 mg/ kg.
3. Experimental Details Failure investigation was conducted on the cracked heater tube. The investigation included visual examination of the tube surfaces and deposits before and after splitting the tube into halves. Also, measurements were done for the variation in the tube wall thickness and deposits. The deposits inside the tube were scrubbed off the surface and analyzed using X-ray diffraction (XRD). Water extract of the deposits was also prepared, and pH measurement was taken. Also, cross-sectional specimens were cut from the tube and metallographically prepared for microhardness measurements, as well as microscopic examinations and analysis of the material. The sections were cut at the crack location and away from the crack. The microhardness measurements (in Vickers) were conducted on a polished crosssection, using diamond pyramid indenter at an applied 2
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
load of 500 g. The measurements were taken at 1-mm intervals along a line starting from the internal surface. The microscopic examinations were carried out before and after etching, using optical and scanning electron microscopy (SEM). Etching was made electrolytically in 10% Oxalic acid at an applied voltage of 6 V. The corrosion products were analyzed, using energy dispersive spectroscopy (EDS). In order to evaluate sensitization of the tube material, and therefore, its susceptibility to intergranular corrosion, full thickness service exposed and solution annealed crosssectional specimens were tested using the double loop electrochemical potentiokinetic reactivation (EPR) technique. The EPR tests were made in accordance with a procedure previously reported [11,12]. The solution annealed specimen was prepared by subjecting a specimen cut from the service exposed tube to a heat treatment at 1050 °C for 1 h followed by water quenching to dissolve the precipitated carbides. The solution for the EPR tests consisted of 0.5-M H2SO4 and 0.01M KSCN. The solution was deaerated by purging with nitrogen gas for 1 h before each test, as well as during the tests, which were conducted at room temperature. A graphite rod and saturated calomel electrode (SCE) were used as the counter and reference electrodes, respectively. Each test was started by keeping the specimen at a cathodic potential of −600 mV vs SCE for 2 min, and then, holding the specimen at its corrosion potential for another 2 min. The EPR plot was obtained through an anodic potential sweep at a scan rate of 100 mV/min starting from the corrosion potential. When the potential reached a value of +300 mV, back scan at the same rate was conducted until the open circuit potential was reached. The ratios of reactivation peak current (Ir) to activation current peak (Ia) of the solution annealed and service exposed specimens were compared to assess the relative degree of sensitization. Both EPR test specimens were examined using the optical microscope and SEM. 4. Results 4.1. Visual Observations The stud tube that suffered cracking in a circumferential direction was visually examined. The crack was found to be not associated with bulging. Thus, the appearance of the crack may suggest that cracking could have been due to stress corrosion caused by polythionic acid or chloride and not to the creep. The provided tube section was 89.0 cm in length and 13.83 cm in outside diameter. The internal tube surface was covered with a deposited black layer (see Fig. 1c and d). The black layer varied considerably in thickness. The minimum thickness of the deposited layer was ≈1.86 mm; while the maximum thickness was ≈4.61 mm (a difference of ≈2.75 mm). The maximum deposit thickness was seen at the crack location. The tube also varied in wall thickness, and the minimum tube thickness was ≈5.92 mm; while the maximum thickness was ≈6.96 mm. Thus, the tube suffered reduction in wall thickness of ≈1.04 mm maximum, which was mostly at the crack location. 4.2. Microhardness Measurements Table 1 provides the average microhardness values taken for a polished crosssection cut from the failed tube. It is clear from the table that the microhardness is slightly higher, close to the internal and external surfaces than at the interior of the crosssection. 4.3. Analysis of Deposits The black deposited layer found on the internal tube surface was analyzed, using XRD. It is clear from Fig. 2 that the black deposited layer consisted of carbon (coke), iron sulphide, calcite, sodium chloride, hematite, and magnetite. It is worth mentioning that the black deposited layer was strongly adherent to the internal metal surface. Moreover, the layer was quite dense and hard, and thus, it was difficult to remove. A chisel was used to remove sufficient amount for XRD. Measurement of pH of the water extract of the deposits yielded an acidic pH of 4.8. 4.4. Electrochemical Reactivation Test The EPR curves obtained for the service exposed and the solution annealed tube material are shown in Fig. 3. Forward scanning was shown to provide passivity at about −150 mV, as indicated by a drop in anodic current. The solution annealed specimen showed a lower passive current compared to that obtained for the service exposed specimen, indicating that the failed tube had less tendency Table 1 Average Microhardness Measurements along Tube Crosssection Starting from Internal Surface. Depth (mm)
Vickers Hardness (Hv)
0–1 1–2 2–3 3–4 4–5
171 164 165 174 183
3
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 2. XRD pattern of black deposited layer found on the internal tube surface.
towards passivation. Upon reverse scanning, reactivation peaks were formed for both specimens, but the reactivation peak current for the service exposed tube was much higher than that of the solution annealed sample. In terms of Ir/Ia%, the service exposed sample showed Ir/Ia% value of 4.0; whereas, the Ir/Ia% value for the solution annealed specimen was 0.07. The relatively high Ir/Ia% value of the service exposed specimen (over 50 times) compared to that obtained for the solution annealed specimen would indicate that the tube material suffered some degree of sensitization during its long-term service. 4.5. Microscopic Examinations and Analyses As previously mentioned, optical and SEM examinations were conducted on the material of the failed tube, as well as, on the specimens that underwent the electrochemical reactivation tests. Also, the alloy base metal, scale, and precipitates were analyzed using EDS. Fig. 4 shows optical micrographs taken for an etched crosssection cut from the tube at the crack location. Shallow grooves or pits
Fig. 3. EPR curves obtained for the service exposed and the solution annealed 321 SS tube material in 0.5-M H2SO4 + 0.01-M KSCN solution.
4
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 4. Optical micrographs of a crosssection cut at the crack location, showing (a) major crack and linear precipitations emanating from its walls close to internal surface; (b) intergranular attack and grain separation at internal surface; (c) appearance of the main crack and branches in the mid tube section; and (d) view of the main crack and nearby feathery cracks close to external surface (oxalic acid etch).
were seen along the internal tube surface underneath a thick scale layer. The grooves appeared to have developed as a result of the separation of internal surface grains (Fig. 4a and b). Apparently, the internal surface experienced grain boundary attack to a depth of two grains, resulting in corrosion and dropping out of surface grains. Several intergranular cracks initiated from the internal tube surface. However, these cracks were not more than four grains in depth, except the main crack that extended across the tube wall thickness. Linear marks of precipitations extended from both sides of the main crack for a short distance along the tube axis (see Fig. 4a). These lines of precipitations were noticed to exist near the internal surface. Some of the grains next to the main crack walls experienced partial or full grain boundary attack to a depth of one or two grains. This grain boundary attack was noticed to have existed until the middle of the tube wall thickness (see Fig. 4a and c). Beyond the middle of the tube crosssection, the main crack was not surrounded with attacked grain boundaries. The main crack had a few short branches, and the crack branches were noticed to be intergranular starting from the internal surface until about half the wall thickness and then became transgranular up to the external surface. Also, feathery transgranular cracks existed near the main crack close to the external surface (Fig. 4d). Subsurface grain boundary precipitations were seen close to the external surface. Optical microscopic examination was also done on a cross-sectional specimen cut away from the crack in order to identify the extent and severity of damage. Once again, subsurface intercrystalline precipitation was observed up to two grains in depth at the internal surface. Also, the internal surface was found to contain small grooves or pits and was covered with a scale layer. The extent of grain boundary precipitation and grain separation appeared considerably less than at the failure location. The linear precipitation close to the internal surface was less intense and much shorter when compared with those observed at the failure location. It is worth mentioning that scattered grain boundary precipitations were seen near the external surface. SEM examinations of a polished crosssection cut at the crack location revealed that the separation of grains, the subsurface grain boundary penetrations, the linear marks, and the crack and its branches were associated with precipitation of a corrosion product compound. The precipitates penetrated into the bulk of the alloy starting from the internal tube surface (Fig. 5). The precipitation appeared similar to those resulting from high temperature reactions, such as oxidation or sulphidation. When the crosssectional specimens that underwent the EPR tests were examined, grain boundary attack manifested by ditches was evident for the service exposed specimen. The grain boundary attack was observed at both the external and internal surfaces, although the depth of attack was much more at the external surface (see Fig. 6). On the other hand, the solution annealed specimen
5
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 5. SEM micrographs of a polished crosssection of the cracked tube, showing (a) scale buildup and corrosion products filling a crack with branches emanating from the internal surface; (b) view of the scale-filled crack and nearby feathery cracks near external surface; and (c) and (d) penetration of boundaries of grains surrounding the major crack near internal surface.
did not suffer grain boundary attack. These results indicated that the internal and external surface layers of the heater tube had suffered sensitization during service. Spot EDS analyses were made on the base metal, internal surface scale, and precipitates inside the crack. As expected, the EDS spectrum of the base metal exhibited peaks for iron (Fe), chromium (Cr), nickel (Ni), and titanium (Ti). On the other hand, EDS analyses of the internal surface scale always provided spectra containing strong Fe and sulphur (S) peaks (Fig. 7a). When the same analysis was made on the precipitates inside the main crack and its branches, the spectrum was most of the time similar to that of the
Fig. 6. SEM micrographs of a crosssection cut from the tube showing intergranular attack at internal surface (a) external surface (b) after EPR test in 0.5-M H2SO4 and 0.01-M KSCN solution.
6
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 7. EDS analysis of (a) internal surface scale and (b) precipitates inside the main crack.
internal surface scale (i.e., Fe and S peaks). But, it occasionally exhibited prominent peaks of Fe, Cr, and S. The Cr peak was noticed to be equal to that of Fe at some spots and higher than that of Fe at other spots (Fig. 7b). These results would suggest that the internal surface scale and the precipitates inside the crack and its branches are mostly iron sulphide. However, the high Cr peak could be due to chromium sulphide precipitations. 5. Discussion The EPR test and the microscopic examination of the tested specimen indicated that the internal and external surface layers of the tube material experienced grain boundary attack. Thus, the tube surfaces were in a sensitized condition, which might explain the slight increase in hardness observed at both surfaces. As previously indicated, the failed tube was in the convection section, where the inlet temperature is 130 °C (266°F); while the outlet temperature is 316 °C (601°F). These temperatures are far below the requirement for the occurrence of sensitization in such a stabilized alloy. However, long-term aging and the deposition of coke and salts from the heavy crude oil during a service that extended more than twenty years are most likely the causes of the observed sensitization. It appeared that the precipitation of coke and salts from the heavy crude led to several events that included sensitization of the tube surface layers and formation of grooves/pits. The presence of chlorides and acidic conditions as indicated by XRD analysis and pH measurement of the deposits autocatalyzed the growth of pits at the internal tube surface. The pits in the sensitized internal surface layer of the tube acted as stress risers to the initiation of chloride stress corrosion cracks. In the sensitized layer close to the internal surface, the cracks propagated intergranularly. The precipitation of coke and salts also led to hot spot formation, which resulted in sulphidation of the internal tube surface and the crack walls. The sulphidation of crack walls obscured the typical features of chloride SCC near the internal surface. However, with the increase in stress concentration accompanying the crack propagation, the crack branches became transgranular in the midsection of the tube, where the alloy was not in a sensitized condition. It is possible that the synergistic action of chloride and H2S played a role in the initiation of stress corrosion cracks. 7
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
The only possible explanation for the occurrence of high temperature sulfidation is the formation of a hot spot along the length of the tube. This view is supported by the fact that the internal tube surface was found covered with a nonuniform black deposited layer. The thickness of the layer was ≈1.86 mm minimum and ≈4.61 mm maximum (a difference of ≈2.75 mm). The maximum deposit thickness was seen near the failure location. In addition, the layer was very dense and hard that a chisel was used to remove sufficient quantity for XRD. This nature of the deposited layer could be attributed to its composition, which consisted of calcite in addition to coke, iron sulfide, sodium chloride, hematite, and magnetite. According to ASM Handbook [13], the presence of internal surface scale or deposits causes an increase in the tube metal temperature, because the scale or deposits have a lower thermal conductivity than the steel tube. Such temperature increase can lead to creep and accelerated oxidation in addition to other failure mechanisms. It would appear that in the present case, the increase in metal temperature did not lead to creep, but caused sensitization and high temperature sulphidation. It has been suggested [13] that for austenitic iron-based high temperature alloys, elevated temperature behavior, such as creep, begins at 0.49Tm (700 °C), where Tm is the melting temperature. Thus, it would seem that the tube was exposed to temperatures greater than 260 °C, but less than 700 °C. Refinery furnace tubes are usually fabricated from several grades of steels for elevated temperature applications, mainly ferritic, i.e., carbon steels, carbon-molybdenum (C-Mo) steels, and Cr-Mo steels. The materials are selected for their stress rupture or creep rupture properties combined with corrosion resistance [14,15]. SSs are sometimes used as furnace tubes for highly corrosive feeds. SSs are preferred over high nickel alloys, because nickel is prone to form low melting nickel-nickel sulphide eutectic. Over the service time, the tubes are subjected to several degradation phenomena that may cause operational problems. Damage to the inside surface of the tube results from the aggressive action of the feed. The principal corrosive impurities in crude oils are sulphur compounds and chlorides [16]. In the present case, chlorides led to SCC, while sulphidic corrosion caused by the sulphur compounds resulted in localized sulphide scale formation on the crack walls and along grain boundaries. As previously mentioned, heavy crude oils rich in salts, sulphur and H2S were the charge since commissioning over twenty years ago. This might explain the presence of calcite as part of the scale and the occurrence of the failure in the convection section where the operating temperature is suitable for the precipitation of calcite. Although minor sulphidation existed along the length of the internal tube surface; albeit, it was more intense at the failure location where more salt precipitation occurred. It can be safely assumed that the nature of the crude was the catalytic factor for the occurrence of SCC and the sulphidation of the internal surface and crack walls. The presence of hematite and magnetite within the scale suggests their formation during decoking operations carried out by means of steam or that the crude oil still contained residual water. The latter assumption appears to be the correct one since not only calcite was precipitated on the internal tube surface, but also sodium chloride. Because metal sulphides form at faster rates than do metal oxides, much more sulphides were observed in the present failure than oxides. Thus, it was advised that the performance of the desalter be improved and frequent decoking and scale removal be carried out, with emphasis on the convection section. 6. Conclusions
• The charge heater tube suffered from chloride SCC. The chloride SCC initiated in the sensitized internal surface layer as intergranular cracks. • The precipitation of salts from the crude increased the metal temperature in the convection section of the heater, leading to •
sensitization of the material and the high temperature sulphidation. Sulphidation of crack walls obscured the chloride SCC features at some locations. The utilization of heavy crude oil rich in salts, sulphur, and H2S may have been the catalytic factor. Thus, the problem is related to the type of feed used.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
The role of stainless steels in petroleum refining, Technical Report 9021. Toronto, Canada: Nickel Institute; 1977. Sedriks AJ. Corrosion of stainless steels. New York: John Wiley and Sons; 1979. Laitinen T. Localized corrosion of stainless steel in chloride, sulfate and thiosulfate containing environments. Corr Sci 2000;42:421–41. Abdulsalam MI, Shinohara T. Technical note: crevice corrosion of stainless steel in hot salt water. Corrosion 2003;59:291–4. Bosch RW. Electrochemical impedance spectroscopy for the detection of stress corrosion cracks in aqueous corrosion systems at ambient and high temperature. Corr Sci 2005;47:125–43. Brophy AJ. Stress corrosion cracking of austentic stainless steels in refinery environments. Mater Performance 1974;13(5):9–15. Gutzeit J, Merrick RD, Scharfstein LR. Corrosion in petroleum refining and petrochemical operations. Metals Handbook 13, Corrosion. 9th ed. Ohio: ASM International Metals Park; 1987. Solomon HD. Transgranular, granulated, and intergranular stress corrosion cracking in AISI 304 SS. Corrosion 1984;40(9):493–506. Singh V. Performance of austenitic stainless steels in wet sour gas-part I. Mater Performance 2004;47(8):52–5. Singh V. Performance of austenitic stainless steels in wet sour gas-part II. Mater Perform 2004;47(9):46–50. Umemura F, Akashi M, Kawamoto T. Evaluation of IGSCC susceptibility of austenitic stainless steels using electrochemical reactivation method, Boshoku Gijutsu. Corros Eng 1980;29(4):163–9. Majidi AP, Streicher MA. The double loop reactivation method for detecting sensitization in AISI 304 stainless steels. Corrosion 1984;40(11):584–93. ASM Handbook, Failure analysis and prevention vol. 11. ASM International; 1986. API 530, Sixth edition, Calculation of heater-tube thickness in petroleum refineries. Washington D.C: API; 2008. White RA, Ehmke EF. Materials Selection for Refineries and Associated Facilities. Houston, Texas: NACE; 1991. p. 131. Gutzeit J. High temperature sulfidic corrosion of steels. Process industry corrosion – the theory and practice. NACE; 1986.
8
Contents lists available at ScienceDirect
Case Studies in Engineering Failure Analysis journal homepage: www.elsevier.com/locate/csefa
Short communication
Failure of 321 stainless steel heater tube in heavy crude oil ⁎
MARK
H.M. Shalaby , K. Ravindranath, N. Tanoli, B. Al-Wakaa Petroleum Research Center, Kuwait Institute for Scientific Research, P.O. Box 24885, 13109-Safat, Kuwait
AR TI CLE I NF O
AB S T R A CT
Keywords: Stress corrosion cracking Charge heaters Refineries Sulphidation Salt deposition
Failure investigation was done on a 321 stainless steel charge heater tube which failed in a refinery unit processing heavy crude oil. Crude oil was the charge in the radiant and convection sections; while saturated stripping steam is present in convection section. After a leak was detected, visual inspection revealed that nine convection tubes had black oil/coke deposits on their external surfaces. The deposits were seen on the first three rows of tubes. When one of the tubes was lightly ground at the black colored area, a circumferential crack was visually observed. The investigation revealed that long-term aging, coupled with localized deposition of salts and coke from the heavy crude led to sensitization of the tube surface layers. This in turn resulted in sulphidation of the internal surface grain boundaries, formation of grooves, and cracking of the material. Thus, cracking was intergranular in nature in the initial stage, but became transgranular at later stages. It was concluded that cracking was due to chloride stress corrosion cracking catalyzed by the presence of sulphur-bearing species. It was recommended that the desalter operation be improved and frequent decoking and scale removal be carried out, with emphasis on the convection section at the refinery.
1. Introduction Stabilized austenitic stainless steels (SSs) have since been widely used in the petroleum refining industry because of their resistance to sulphidic corrosion and polythionic acid stress corrosion cracking (SCC), as well as of their excellent strength and toughness [1,2]. However, in some instances, stabilized austenitic SSs have been found to be susceptible to pitting, crevice corrosion, and SCC [3–5]. The susceptibility to SCC occurs under conditions involving high stress, changes in metallurgical structure due to high temperature exposure, and the presence of specific chemicals that promote cracking [6]. It was found that the factors most affecting corrosion of SSs in the refinery industry are chloride and hydrogen sulphide [3]. Hydrogen sulphide is one of the constituents of crude oils and refinery sour waters and is also formed by the decomposition of organic sulfur compounds at elevated temperatures [7]. Chloride SCC of austenitic SSs is transgranular, but can be intergranular if the material is sensitized [8]. There are reported failure cases suggesting that the presence of hydrogen sulphide (H2S) synergistically promotes chloride SCC [9,10]. Recent changes in refinery feedstocks in terms of increased H2S content have accelerated corrosion failures of stabilized austenitic SSs equipment. The present paper presents an example of such failures, where a 321 SS charge heater tube carrying heavy sour crude oil suffered cracking. 2. Background A leak occurred in a heater of a refinery unit handling heavy sour crude oil. Visual inspection revealed that eight convection tubes
⁎
Corresponding author. E-mail address: [email protected] (H.M. Shalaby).
http://dx.doi.org/10.1016/j.csefa.2017.04.004 Received 26 March 2017; Accepted 28 April 2017 Available online 20 May 2017 2213-2902/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 1. Camera photographs showing (a) as-received tube section; (b) circumferential crack seen after removal of studs; (c) deposited black layer on internal surface; and (d) stud portion of tube.
made of 321 SS had black oil/coke deposits on their external surfaces. The deposits were seen on the first two rows of plain tubes. The observations were made through available gaps between tubes, as access to the convection tubes was severely limited. In order to identify the leaky tubes, the refinery engineers pressurized the heater. However, no leak was detected as seen from the radiant zone. The eight plain tubes having black deposits were removed from the heater, which facilitated inspection of stud tubes. One stud tube in the third row was found to have black deposits. This tube was also removed, and the removed nine tubes were replaced with new ones. The removed 321 SS tubes were examined using dye penetrant test. No indication was found on all tubes. However, when the stud tube was lightly grounded externally at the black colored area, a circumferential crack was visually observed (Fig. 1a and b). The presence of this crack raised the possibility that the remaining heater tubes in operation may also have cracks. The heater is part of a heavy crude topping unit. The crude was the charge (tube side) in the radiant and convection sections since its commissioning over twenty years ago. Saturated stripping steam is present in the convection section. The temperatures at the inlet radiant and convection are 130 °C (266°F). On the other hand, the outlet temperatures are 345 °C (653°F) and 316 °C (601°F). The heavy crude charge has gravity in the order of 17 American Petroleum Institute (API), a total sulphur content of 4.48% wt, and organic chloride content of 3 mg/kg. However, this crude was changed to another heavy crude oil, a year earlier than the date of the current failure. This latter crude has 20.48 API gravity, a total sulphur content of 3.91% wt, and organic chloride content of 1 mg/ kg.
3. Experimental Details Failure investigation was conducted on the cracked heater tube. The investigation included visual examination of the tube surfaces and deposits before and after splitting the tube into halves. Also, measurements were done for the variation in the tube wall thickness and deposits. The deposits inside the tube were scrubbed off the surface and analyzed using X-ray diffraction (XRD). Water extract of the deposits was also prepared, and pH measurement was taken. Also, cross-sectional specimens were cut from the tube and metallographically prepared for microhardness measurements, as well as microscopic examinations and analysis of the material. The sections were cut at the crack location and away from the crack. The microhardness measurements (in Vickers) were conducted on a polished crosssection, using diamond pyramid indenter at an applied 2
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
load of 500 g. The measurements were taken at 1-mm intervals along a line starting from the internal surface. The microscopic examinations were carried out before and after etching, using optical and scanning electron microscopy (SEM). Etching was made electrolytically in 10% Oxalic acid at an applied voltage of 6 V. The corrosion products were analyzed, using energy dispersive spectroscopy (EDS). In order to evaluate sensitization of the tube material, and therefore, its susceptibility to intergranular corrosion, full thickness service exposed and solution annealed crosssectional specimens were tested using the double loop electrochemical potentiokinetic reactivation (EPR) technique. The EPR tests were made in accordance with a procedure previously reported [11,12]. The solution annealed specimen was prepared by subjecting a specimen cut from the service exposed tube to a heat treatment at 1050 °C for 1 h followed by water quenching to dissolve the precipitated carbides. The solution for the EPR tests consisted of 0.5-M H2SO4 and 0.01M KSCN. The solution was deaerated by purging with nitrogen gas for 1 h before each test, as well as during the tests, which were conducted at room temperature. A graphite rod and saturated calomel electrode (SCE) were used as the counter and reference electrodes, respectively. Each test was started by keeping the specimen at a cathodic potential of −600 mV vs SCE for 2 min, and then, holding the specimen at its corrosion potential for another 2 min. The EPR plot was obtained through an anodic potential sweep at a scan rate of 100 mV/min starting from the corrosion potential. When the potential reached a value of +300 mV, back scan at the same rate was conducted until the open circuit potential was reached. The ratios of reactivation peak current (Ir) to activation current peak (Ia) of the solution annealed and service exposed specimens were compared to assess the relative degree of sensitization. Both EPR test specimens were examined using the optical microscope and SEM. 4. Results 4.1. Visual Observations The stud tube that suffered cracking in a circumferential direction was visually examined. The crack was found to be not associated with bulging. Thus, the appearance of the crack may suggest that cracking could have been due to stress corrosion caused by polythionic acid or chloride and not to the creep. The provided tube section was 89.0 cm in length and 13.83 cm in outside diameter. The internal tube surface was covered with a deposited black layer (see Fig. 1c and d). The black layer varied considerably in thickness. The minimum thickness of the deposited layer was ≈1.86 mm; while the maximum thickness was ≈4.61 mm (a difference of ≈2.75 mm). The maximum deposit thickness was seen at the crack location. The tube also varied in wall thickness, and the minimum tube thickness was ≈5.92 mm; while the maximum thickness was ≈6.96 mm. Thus, the tube suffered reduction in wall thickness of ≈1.04 mm maximum, which was mostly at the crack location. 4.2. Microhardness Measurements Table 1 provides the average microhardness values taken for a polished crosssection cut from the failed tube. It is clear from the table that the microhardness is slightly higher, close to the internal and external surfaces than at the interior of the crosssection. 4.3. Analysis of Deposits The black deposited layer found on the internal tube surface was analyzed, using XRD. It is clear from Fig. 2 that the black deposited layer consisted of carbon (coke), iron sulphide, calcite, sodium chloride, hematite, and magnetite. It is worth mentioning that the black deposited layer was strongly adherent to the internal metal surface. Moreover, the layer was quite dense and hard, and thus, it was difficult to remove. A chisel was used to remove sufficient amount for XRD. Measurement of pH of the water extract of the deposits yielded an acidic pH of 4.8. 4.4. Electrochemical Reactivation Test The EPR curves obtained for the service exposed and the solution annealed tube material are shown in Fig. 3. Forward scanning was shown to provide passivity at about −150 mV, as indicated by a drop in anodic current. The solution annealed specimen showed a lower passive current compared to that obtained for the service exposed specimen, indicating that the failed tube had less tendency Table 1 Average Microhardness Measurements along Tube Crosssection Starting from Internal Surface. Depth (mm)
Vickers Hardness (Hv)
0–1 1–2 2–3 3–4 4–5
171 164 165 174 183
3
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 2. XRD pattern of black deposited layer found on the internal tube surface.
towards passivation. Upon reverse scanning, reactivation peaks were formed for both specimens, but the reactivation peak current for the service exposed tube was much higher than that of the solution annealed sample. In terms of Ir/Ia%, the service exposed sample showed Ir/Ia% value of 4.0; whereas, the Ir/Ia% value for the solution annealed specimen was 0.07. The relatively high Ir/Ia% value of the service exposed specimen (over 50 times) compared to that obtained for the solution annealed specimen would indicate that the tube material suffered some degree of sensitization during its long-term service. 4.5. Microscopic Examinations and Analyses As previously mentioned, optical and SEM examinations were conducted on the material of the failed tube, as well as, on the specimens that underwent the electrochemical reactivation tests. Also, the alloy base metal, scale, and precipitates were analyzed using EDS. Fig. 4 shows optical micrographs taken for an etched crosssection cut from the tube at the crack location. Shallow grooves or pits
Fig. 3. EPR curves obtained for the service exposed and the solution annealed 321 SS tube material in 0.5-M H2SO4 + 0.01-M KSCN solution.
4
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 4. Optical micrographs of a crosssection cut at the crack location, showing (a) major crack and linear precipitations emanating from its walls close to internal surface; (b) intergranular attack and grain separation at internal surface; (c) appearance of the main crack and branches in the mid tube section; and (d) view of the main crack and nearby feathery cracks close to external surface (oxalic acid etch).
were seen along the internal tube surface underneath a thick scale layer. The grooves appeared to have developed as a result of the separation of internal surface grains (Fig. 4a and b). Apparently, the internal surface experienced grain boundary attack to a depth of two grains, resulting in corrosion and dropping out of surface grains. Several intergranular cracks initiated from the internal tube surface. However, these cracks were not more than four grains in depth, except the main crack that extended across the tube wall thickness. Linear marks of precipitations extended from both sides of the main crack for a short distance along the tube axis (see Fig. 4a). These lines of precipitations were noticed to exist near the internal surface. Some of the grains next to the main crack walls experienced partial or full grain boundary attack to a depth of one or two grains. This grain boundary attack was noticed to have existed until the middle of the tube wall thickness (see Fig. 4a and c). Beyond the middle of the tube crosssection, the main crack was not surrounded with attacked grain boundaries. The main crack had a few short branches, and the crack branches were noticed to be intergranular starting from the internal surface until about half the wall thickness and then became transgranular up to the external surface. Also, feathery transgranular cracks existed near the main crack close to the external surface (Fig. 4d). Subsurface grain boundary precipitations were seen close to the external surface. Optical microscopic examination was also done on a cross-sectional specimen cut away from the crack in order to identify the extent and severity of damage. Once again, subsurface intercrystalline precipitation was observed up to two grains in depth at the internal surface. Also, the internal surface was found to contain small grooves or pits and was covered with a scale layer. The extent of grain boundary precipitation and grain separation appeared considerably less than at the failure location. The linear precipitation close to the internal surface was less intense and much shorter when compared with those observed at the failure location. It is worth mentioning that scattered grain boundary precipitations were seen near the external surface. SEM examinations of a polished crosssection cut at the crack location revealed that the separation of grains, the subsurface grain boundary penetrations, the linear marks, and the crack and its branches were associated with precipitation of a corrosion product compound. The precipitates penetrated into the bulk of the alloy starting from the internal tube surface (Fig. 5). The precipitation appeared similar to those resulting from high temperature reactions, such as oxidation or sulphidation. When the crosssectional specimens that underwent the EPR tests were examined, grain boundary attack manifested by ditches was evident for the service exposed specimen. The grain boundary attack was observed at both the external and internal surfaces, although the depth of attack was much more at the external surface (see Fig. 6). On the other hand, the solution annealed specimen
5
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 5. SEM micrographs of a polished crosssection of the cracked tube, showing (a) scale buildup and corrosion products filling a crack with branches emanating from the internal surface; (b) view of the scale-filled crack and nearby feathery cracks near external surface; and (c) and (d) penetration of boundaries of grains surrounding the major crack near internal surface.
did not suffer grain boundary attack. These results indicated that the internal and external surface layers of the heater tube had suffered sensitization during service. Spot EDS analyses were made on the base metal, internal surface scale, and precipitates inside the crack. As expected, the EDS spectrum of the base metal exhibited peaks for iron (Fe), chromium (Cr), nickel (Ni), and titanium (Ti). On the other hand, EDS analyses of the internal surface scale always provided spectra containing strong Fe and sulphur (S) peaks (Fig. 7a). When the same analysis was made on the precipitates inside the main crack and its branches, the spectrum was most of the time similar to that of the
Fig. 6. SEM micrographs of a crosssection cut from the tube showing intergranular attack at internal surface (a) external surface (b) after EPR test in 0.5-M H2SO4 and 0.01-M KSCN solution.
6
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
Fig. 7. EDS analysis of (a) internal surface scale and (b) precipitates inside the main crack.
internal surface scale (i.e., Fe and S peaks). But, it occasionally exhibited prominent peaks of Fe, Cr, and S. The Cr peak was noticed to be equal to that of Fe at some spots and higher than that of Fe at other spots (Fig. 7b). These results would suggest that the internal surface scale and the precipitates inside the crack and its branches are mostly iron sulphide. However, the high Cr peak could be due to chromium sulphide precipitations. 5. Discussion The EPR test and the microscopic examination of the tested specimen indicated that the internal and external surface layers of the tube material experienced grain boundary attack. Thus, the tube surfaces were in a sensitized condition, which might explain the slight increase in hardness observed at both surfaces. As previously indicated, the failed tube was in the convection section, where the inlet temperature is 130 °C (266°F); while the outlet temperature is 316 °C (601°F). These temperatures are far below the requirement for the occurrence of sensitization in such a stabilized alloy. However, long-term aging and the deposition of coke and salts from the heavy crude oil during a service that extended more than twenty years are most likely the causes of the observed sensitization. It appeared that the precipitation of coke and salts from the heavy crude led to several events that included sensitization of the tube surface layers and formation of grooves/pits. The presence of chlorides and acidic conditions as indicated by XRD analysis and pH measurement of the deposits autocatalyzed the growth of pits at the internal tube surface. The pits in the sensitized internal surface layer of the tube acted as stress risers to the initiation of chloride stress corrosion cracks. In the sensitized layer close to the internal surface, the cracks propagated intergranularly. The precipitation of coke and salts also led to hot spot formation, which resulted in sulphidation of the internal tube surface and the crack walls. The sulphidation of crack walls obscured the typical features of chloride SCC near the internal surface. However, with the increase in stress concentration accompanying the crack propagation, the crack branches became transgranular in the midsection of the tube, where the alloy was not in a sensitized condition. It is possible that the synergistic action of chloride and H2S played a role in the initiation of stress corrosion cracks. 7
Case Studies in Engineering Failure Analysis 9 (2017) 1–8
H.M. Shalaby et al.
The only possible explanation for the occurrence of high temperature sulfidation is the formation of a hot spot along the length of the tube. This view is supported by the fact that the internal tube surface was found covered with a nonuniform black deposited layer. The thickness of the layer was ≈1.86 mm minimum and ≈4.61 mm maximum (a difference of ≈2.75 mm). The maximum deposit thickness was seen near the failure location. In addition, the layer was very dense and hard that a chisel was used to remove sufficient quantity for XRD. This nature of the deposited layer could be attributed to its composition, which consisted of calcite in addition to coke, iron sulfide, sodium chloride, hematite, and magnetite. According to ASM Handbook [13], the presence of internal surface scale or deposits causes an increase in the tube metal temperature, because the scale or deposits have a lower thermal conductivity than the steel tube. Such temperature increase can lead to creep and accelerated oxidation in addition to other failure mechanisms. It would appear that in the present case, the increase in metal temperature did not lead to creep, but caused sensitization and high temperature sulphidation. It has been suggested [13] that for austenitic iron-based high temperature alloys, elevated temperature behavior, such as creep, begins at 0.49Tm (700 °C), where Tm is the melting temperature. Thus, it would seem that the tube was exposed to temperatures greater than 260 °C, but less than 700 °C. Refinery furnace tubes are usually fabricated from several grades of steels for elevated temperature applications, mainly ferritic, i.e., carbon steels, carbon-molybdenum (C-Mo) steels, and Cr-Mo steels. The materials are selected for their stress rupture or creep rupture properties combined with corrosion resistance [14,15]. SSs are sometimes used as furnace tubes for highly corrosive feeds. SSs are preferred over high nickel alloys, because nickel is prone to form low melting nickel-nickel sulphide eutectic. Over the service time, the tubes are subjected to several degradation phenomena that may cause operational problems. Damage to the inside surface of the tube results from the aggressive action of the feed. The principal corrosive impurities in crude oils are sulphur compounds and chlorides [16]. In the present case, chlorides led to SCC, while sulphidic corrosion caused by the sulphur compounds resulted in localized sulphide scale formation on the crack walls and along grain boundaries. As previously mentioned, heavy crude oils rich in salts, sulphur and H2S were the charge since commissioning over twenty years ago. This might explain the presence of calcite as part of the scale and the occurrence of the failure in the convection section where the operating temperature is suitable for the precipitation of calcite. Although minor sulphidation existed along the length of the internal tube surface; albeit, it was more intense at the failure location where more salt precipitation occurred. It can be safely assumed that the nature of the crude was the catalytic factor for the occurrence of SCC and the sulphidation of the internal surface and crack walls. The presence of hematite and magnetite within the scale suggests their formation during decoking operations carried out by means of steam or that the crude oil still contained residual water. The latter assumption appears to be the correct one since not only calcite was precipitated on the internal tube surface, but also sodium chloride. Because metal sulphides form at faster rates than do metal oxides, much more sulphides were observed in the present failure than oxides. Thus, it was advised that the performance of the desalter be improved and frequent decoking and scale removal be carried out, with emphasis on the convection section. 6. Conclusions
• The charge heater tube suffered from chloride SCC. The chloride SCC initiated in the sensitized internal surface layer as intergranular cracks. • The precipitation of salts from the crude increased the metal temperature in the convection section of the heater, leading to •
sensitization of the material and the high temperature sulphidation. Sulphidation of crack walls obscured the chloride SCC features at some locations. The utilization of heavy crude oil rich in salts, sulphur, and H2S may have been the catalytic factor. Thus, the problem is related to the type of feed used.
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
The role of stainless steels in petroleum refining, Technical Report 9021. Toronto, Canada: Nickel Institute; 1977. Sedriks AJ. Corrosion of stainless steels. New York: John Wiley and Sons; 1979. Laitinen T. Localized corrosion of stainless steel in chloride, sulfate and thiosulfate containing environments. Corr Sci 2000;42:421–41. Abdulsalam MI, Shinohara T. Technical note: crevice corrosion of stainless steel in hot salt water. Corrosion 2003;59:291–4. Bosch RW. Electrochemical impedance spectroscopy for the detection of stress corrosion cracks in aqueous corrosion systems at ambient and high temperature. Corr Sci 2005;47:125–43. Brophy AJ. Stress corrosion cracking of austentic stainless steels in refinery environments. Mater Performance 1974;13(5):9–15. Gutzeit J, Merrick RD, Scharfstein LR. Corrosion in petroleum refining and petrochemical operations. Metals Handbook 13, Corrosion. 9th ed. Ohio: ASM International Metals Park; 1987. Solomon HD. Transgranular, granulated, and intergranular stress corrosion cracking in AISI 304 SS. Corrosion 1984;40(9):493–506. Singh V. Performance of austenitic stainless steels in wet sour gas-part I. Mater Performance 2004;47(8):52–5. Singh V. Performance of austenitic stainless steels in wet sour gas-part II. Mater Perform 2004;47(9):46–50. Umemura F, Akashi M, Kawamoto T. Evaluation of IGSCC susceptibility of austenitic stainless steels using electrochemical reactivation method, Boshoku Gijutsu. Corros Eng 1980;29(4):163–9. Majidi AP, Streicher MA. The double loop reactivation method for detecting sensitization in AISI 304 stainless steels. Corrosion 1984;40(11):584–93. ASM Handbook, Failure analysis and prevention vol. 11. ASM International; 1986. API 530, Sixth edition, Calculation of heater-tube thickness in petroleum refineries. Washington D.C: API; 2008. White RA, Ehmke EF. Materials Selection for Refineries and Associated Facilities. Houston, Texas: NACE; 1991. p. 131. Gutzeit J. High temperature sulfidic corrosion of steels. Process industry corrosion – the theory and practice. NACE; 1986.
8