J Fail. Anal. and Preven. (2010) 10:480 485 DOI 10.1007/s11668-010-9389-9 Failure Analysis of Cracked Reducer Flange Khaled Habib Submitted: 18 May 2010/in revised form: 29 July 2010/Published online: 5 October 2010 ASM International 2010 Abstract A cracked reducer flange was analyzed for the cause of the failure. The flange was carefully cut to obtain samples for metallographic, X-ray, and scanning electron microscopy (SEM) Examinations. The examinations revealed that the introduction of chloride ions in the operational service led to pitting corrosion in the inner surface of the flange. Chloride ion inclusions were probably the result of chemical contaminations, i.e., cleaning chemicals contamination during shutdown of the operation. The introduction of corrosion pits caused unexpected load stress intensification and cracking of the flange. Consequently, stress corrosion cracking emanated from the pits under the influence of chloride attack and operational pressure. Then the cracks propagated in a transgranular manner, in the radial direction of the flange, until the final failure occurred. Keywords Reducer flange Pitting corrosion Stress corrosion cracking Metallographic examination X-ray Scanning electron microscopy Fracture roughness Stress intensity factor (1.5 00 ), of a reactor outlet (V41-002) drain line operating in a refinery plant of the company since 1984, see Fig. 1 for the location of the flange with respect to the (V41-002) reactor in the refinery of the company. The flange made of an austenitic stainless steel, UNSS No. 321, with chemical composition of 0.08% C, 2.0% Mn, 1.0% Si, 17.0 19.0% Cr, 9.0 12.0% Ni, 0.045% P, 0.03% S, minimum 5% Ti, and the balance Fe [1]. The operational pressure and temperature of the drain line of the reactor outlet were 1907 Psig (13.2 MPa) and 763 F (406 C), respectively. On the other hand, the design pressure and temperature of the drain line of the reactor outlet were 2240 Psig (15.5 MPa) and 800 F (426.6 C), respectively. The reactor feeds the drain line with hydrocarbons of H 2, H 2S, H 2O, and NH 3. Also, it has been reported that the partial pressure of the hydrogen was 1600 Psig (11.0 Mpa) in service as compared to the design partial pressure of 1750 Psig (12.2 MPa). In addition, leaks of hydrocarbons were observed which led to the shutdown of the reactor. Experimental Details Visual Inspection Introduction A cracked flange was submitted to Kuwait institute for scientific research (KISR) by a national petroleum company in Kuwait for determination of the cause of the failure. The cracked flange was a reducer flange, a 10 cm (4 00 ) 9 3.75 cm K. Habib (&) A visual inspection of the failed flange indicated that the flange had a primary crack in the radial direction of the flange; Fig. 2 shows the precise site of the crack. Also, there was no indication that plastic deformation took place during the failure processes of the flange. In other words, a brittle fracture was the mode of failure in this case. Materials Science Lab, Department of Advanced Systems, KISR, P.O. Box 24885, 13109 Safat, Kuwait e-mail: khaledhabib@usa.net Metallographic Examinations 123
J Fail. Anal. and Preven. (2010) 10:480 485 481 In order to examine the microstructures of the material and the fracture surface of the primary crack, the flange was cut Fig. 1 A schematic drawing of reactors and connections of the refinery of a petroleum company in Kuwait. The arrow indicates the location of the reducer flange in the (V41-002) reactor in the refinery Fig. 2 A primary crack in the radial direction of the flange
with especial care in the radial direction first to separate the bore of the flange from main body of the flange; see Fig. 3 for the radial cut of the bore of the flange. Then, the bore was cut in the longitudinal direction of the flange to facilitate the opening of the mouth of the primary crack for examinations in a scanning electron microscope (SEM); Fig. 4 shows the longitudinal cut of the bore of the flange. Figure 5 shows the flange after cutting the bore, where the crack is clearly emanating in radial direction from inside the bore toward to the main body of the flange; see the arrow in Fig. 5. Figure 6 represents a microstructure of the material of the Fig. 3 A top view of the radial cut of the pore of the flange failed flange, from the top portion of the sample in Fig. 3, near the main crack. Figure 6 represents a typical microstructure of the austenitic stainless steel of UNSG No. 321 stainless steel, with second phase particles, mainly composed of precipitates of Ti C, precipitated as a 123
J Fail. Anal. and Preven. (2010) 10:480 485 483 482 J Fail. Anal. and Preven. (2010) 10:480 485 Fig. 4 A top view of the longitudinal cut of the pore of the flange grains(a), secondary particles, and annealing twins, 2009 methanolic aqua regia Fig. 5 The view of a flange after cutting the pore, where the crack is clearly emanating in radial direction from inside the pore toward the main body of the flange Fig. 7 Representation of another microstructure of the material from the same portion of the sample in Fig. 3, in which secondary cracks were observed to initiate from the surface of the sample in Fig. 3 near the main crack; then the cracks were observed to branch inside the bulk of the material Fig. 6 Representation of a microstructure of the material of the failed flange, from the top portion of the sample in Fig. 3, near the main crack. The microstructure of the stainless steel contains austenitic Fig. 8 Representation of a microstructure of the failed flange from the inner side of the sample in Fig. 3, near the main crack
result of the aging processes of the austenitic, solid solution, phase in high temperature and high pressure environments. Also, the microstructure contains some annealing twins. These observations of the stainless steel microstructures are in agreement with similar observations on the same stainless steel in the literature [2]. Figure 7 represents another microstructure of the material from the same portion of the sample in Fig. 3, in which secondary cracks were observed to initiate from the surface of the sample in Fig. 3 near the main crack, and then they were observed to branch inside toward the bulk of the material. Figure 8 represents a microstructure of the failed flange from the inner side of the sample in Fig. 3, near the main crack. This figure contains secondary cracks from the inner surface of the flange. The cracks seem to propagate from a second phase particle to another in a transgranular manner, even though the sample has not been chemically etched. Some of the second phase particles were ruptured and desponded from the surface of the microstructure. There was no indication of sensitization in the microstructure of the flange. This can be interpreted to the presence of Ti element which prevented sensitization of the microstructure of the austenitic stainless steel. SEM Examinations and X-Ray Micro-Analysis SEM examinations of the fracture surface of the flange were carried out in the SEM to characterize the mode of the fracture. Figure 9 shows a fractograph of the fractured surface of the flange. It is quite clear from Fig. 9 that the fractograph has a cleavage feature covered partially with corrosion products. An energy despersive spectroscopy (EDS) was conducted on the fractured surface of the stainless steel to determine the chemical elements of the stainless steel sample. An elemental probe microanalysis (EPMA) attached to the SEM microscope was used to determine the EDS spectra near the surface of the stainless steel. Figure 10 exhibits the EDS spectra of the chemical elements contained in Fig. 9, in the stainless steel. The spectra comprise those elements which normally are found in the UNSS No. 321 Stainless steel, except for the sulfur element which is found to be more than usual, because sulfur is only a trace element in this particular stainless steel; see above for the chemical composition of the stainless steel [1]. As a result, one may suggest that H 2S had a role only in the chemical attack of the stainless steel due to the leakage of H 2S into the bore of the flange after the failure of the flange took place, but not in the mechanism to cause the failure. Figure 11 shows a SEM photograph from the inner side of the sample of Fig. 3. Fig. 9 The view of the fractured surface of the flange The photograph exhibits clearly secondary cracks emanating from corrosion pits, in which some corrosion products remain inside these pits. Figure 12 represents EDS spectra of the chemical analysis of the corresponding material shown in Fig. 11. The spectra consist of chemical elements which normally are found in the UNSG No. 321 stainless steel, except chloride ion. This radical element may be introduced into the inside of the flange and may be attributed to chemical contaminations, i.e., cleaning chemical s contamination during shutdown. It is well known that austenitic stainless steels are susceptible to pitting corrosion in the presence of chloride ion Fig. 10 The EDS spectra of the chemical elements in Fig. 9, in the stainless steel 123
J Fail. Anal. and Preven. (2010) 10:480 485 485 Fig. 11 A SEM photograph from the inner side of the sample of Fig. 3 Fig. 13 Another SEM photograph from the inner side of the sample in Fig. 3 solutions [3]. The presence of chloride ions causes localized interruption to the passive film on the stainless steel inducing localized perforation at the metal surface, similar to those observed in Fig. 11. Figure 13 represents another SEM photograph from the inner side of the sample in Fig. 3. The photograph also contains secondary cracks emanating from corrosion pits similar to those seen in Fig. 11. In fact, the presence of secondary cracks initiating from corrosion pits were observed in several locations at the inner surface the flange. Hardness Test A number of Rockwell hardness tests, scale (HRB), were conducted on the stainless steel sample. The hardness Fig. 14 The hardness values imposed on the stainless steel sample values were found to range from 73 to 91.3 in the radial 484 J Fail. Anal. and Preven. (2010) 10:480 485 Fig. 12 The EDS spectra of the chemical analysis of Fig. 11 direction toward the inner surface of the bore of the flange; see Fig. 14 for the range of the hardness values imposed on
the stainless steel sample. From the range of the values of the hardness tests, it is obvious that the stainless steel had not suffered from embrittlement during the 14 years of service, based on the maximum HRB = 95 the published value for the UNSG No. 321 [1] in the literature. The average value of the hardness tests was determined to be HRB = 81 based on an average of five tests. It is known that the austenitic stainless steel, FCC structure, is not susceptible to hydrogen embrittlement in H 2S environments, because of the high content of corrosion-resistant elements, i.e., Cr and Ni, in such a stainless steel [4]. Results and Discussion connected tube of the flange. Therefore, from the tangential stress of the connected tube [5] r t ¼ P i½ðb 2 þ a 2 Þ=b 2 a 2 ðeq 1Þ where r t is the tangential stress; P i is the internal pressure, 15.5 MPa; a is the inner radius of the connected tube; and b is the outer radius of the connected tube. From the longitudinal stress of the connected tube [5] r l ¼ P i ðeq 2Þ where r l is the longitudinal stress. One can calculate the maximum shear stress, s max, from a design philosophy against yielding as the follows [5]: Mechanism of Failure s max ¼ f½ðr t r lþ=2 2 g 0:5 ðeq 3Þ It is clear now that the introduction of chloride ions in the operational service of the flange has led to pitting corrosion to take place in the inner surface of the flange, see Figs. 11 13. The presence of chloride ions was probably due to chemical contaminations, i.e., cleaning chemical s contamination during shutdown of the reactor. Because the flange was operating as a part of a thick-walled high pressure vessel, the flange was under radial and tangential operational stresses, which should not exceed the maximum stress rating of the flange material. However, it was the introduction of the corrosion pits, unexpected notches, into the inner surface of the flange, which led to cracking of the material of the flange. Consequently, stress corrosion cracking emanated from the pits under the influence of chloride attack and operational pressure; then, the cracks propagated from a second phase particle to another in a transgranular manner, in the radial direction of the flange, until the final failure occurred. As a result, a leakage of the hydrocarbons took place, and sulfide attack on the fracture surface occurred. It is worth noting that a similar observation was documented in the literature on an austenitic stainless steel, Type UNSG No. 321, in a similar environment [4]. Mathematical Model In order to describe the way in which the flange failed, from a design philosophy and fracture mechanics point of views, the following analysis has been considered. The failed flange is considered as a part of a thick-walled pressure vessel [5] subjected to not only high temperature but also to internal tangential and longitudinal stresses. This consideration is based on the assumption that the thickness of the flange is more than 0.1 of the internal radius of the We assumed in this failure case that the design of the flange was originally adequate to withstand yielding, permanent deformation. In other words, the calculated maximum shear stress must not exceed at least one half of the yield strength of the materials of the flange, r y [5] according to the maximum shear stress theory. However, unfortunately, the flange failed because of the introduction of the corrosion pits, unexpected notches, to the inner surface of the flange, which led to lower the critical fracture toughness of the threshold of stress corrosion cracking of the material of the flange. Consequently, stress corrosion cracking emanated from the pits under the influence of chloride attack and operational pressure, and then the cracks propagated from a second phase particle to another in a transgranular manner, in the radial direction of the flange, until the final failure occurred. As a result, a leakage of the hydrocarbons took place and sulfide attack on the fracture surface occurred. This implies that the mechanism of the flange failure was due to an unexpected stress raiser that has not been accounted for during the initial design of the flange against brittle fracture. This means that the maximum principal stress of the flange given by r max ¼ ðr t þ r lþ=2 þ s max ðeq 4Þ must not exceed the ultimate strength of the materials of the flange, r max, while taking into account a stress intensity factor, K I, from fracture mechanics design point of view [6], against brittle fracture like the stress corrosion cracking in the failed flange. In other words, the design criterion from fracture mechanics point of view is as follows: ðk Ir maxþ=ðr ultþ\0:5 ðeq 5Þ where K I can be calculated based on the size and the shape of the expected localized damages, i.e., pits or cracks [6]. 123
J Fail. Anal. and Preven. (2010) 10:480 485 487 Conclusions The following conclusions are drawn from the above investigation: (1) A cracked reducer flange was examined to analyze the cause of the failure. The examinations revealed that the introduction of chloride ions in the operational service led to pitting corrosion in the inner surface of the flange. (2) Chloride ions were probably the result of chemical contaminations, i.e., cleaning chemical s contamination during shutdown of the operation. (3) The introduction of corrosion pits caused unexpected load stress intensification and cracking of the flange. Consequently, stress corrosion cracking emanated from the pits under the influence of chloride attack and operational pressure. Then the cracks propagated in a transgranular manner, in the radial direction of the flange, until the final failure occurred. References 1. American Society for Metals (ASM): Metals Handbook, Propertiesand Selection, Iron, Steel, and High Performance Alloys, vol. 1, 10th edn, pp. 843, 856. American Society for Metals, Materials Park, OH (1990) 2. American Society for Metals (ASM): ASM Standards in BuildingCodes, vol. 1,A1-B210 M, 13th edn, p. 508. American Society for Metals, Materials Park, OH (1993) 3. Uhlig, H.: Corrosion and Corrosion Control, 3rd edn, p. 36. JohnWiley & Sons Inc., New york (1971) 4. American Society for Metals (ASM): Case Histories in FailureAnalysis, p. 366. American Society for Metals, Materials Park, OH (1979) 5. Shigley, J.W.: Mechanical Engineering Design, 3rd edn, pp. 28,60, 169. McGraw-Hill, New York (1977) 6. Murakami, Y.: Stress Intensity Factors Handbook, vol. 1, p. 307. Pergamon, Oxford (1987)