Failure Analysis of Pressure Vessel
Failure Analysis of Pressure Vessel
Failure Analysis of Pressure Vessel
Kuisheng Wang
Beijing University of Chemical Technology, Beijing 100029, China
Liping Gao
Hebei Engineering Institute, Handan 056027, China
Introduction
Hydrogen oxygen generating system by water electrolysis is widely used to supply hydrogen in industries such as petroleum, chemistry, metallurgy and electronics, etc., and supply oxygen for submarines and closed systems. The oxygen separator is an important pressure vessel in the system. In recent two years, we have collected dozens of reports about the corrosion failure of oxygen separator. Some common features and conditions of them are as follows: i. All cracks happen on the oxygen separator while nothing occurs in the hydrogen separator. ii. The position and direction of cracks are basically similar, they are located at the straight edges of the heads. The direction of them is vertical to the round weld and their distance from the round weld is about 4 10 mm. However, there is no crack happening in itself. iii. Electrolyte used: alkaline solution e.g., 25 30 percent KOH with over-saturated oxygen. iv. Operation temperature: 70 80C; Operation pressure: 1.5 Mpa. v. Duration of operation before failure is about one to two years, but in some cases, the failure takes place just in three months. vi. Technical dimensions: 500 6, 600 10, 800 10. The used material is 1Cr18Ni9Ti in most cases, also SUS 304 and 316L, etc.
2.2 Metallographic Inspection on the Vertical Section The cracks on the vertical section are shown in Fig. 4, in which an obvious dendriform branch is found. The material structure is of austenite, deformed martensite and ferrite Fig. 5 and branches at the tips and transcrystallization are found in some area shown in Fig. 6. The martensite in material structure is formed because of the cold-rolling process of enveloping heads of oxygen separator, and the ferrite is resulted from heating during welding two parts together enveloping heads and cylinder of oxygen separator. There are impurities in the material that have no relation with the formation and extension of cracks.
With the aid of D-max/rb X-ray diffractometer made by the Physical-chemical Electrics Company of Japan with a CO target and a graphite single-color device at a scanning speed of 6 0 /min, an analysis has been done on the composition of the corrosion products. The result of analysis shows that the base is of two phases and the corrosion compositions are Ca3 AlF(SO4 )(OH) 8 and Fe3 O4 , shown in Fig. 13. Transactions of the ASME
Fig. 4 The cracks on the vertical section have obvious dendriform branch
Discussion
Based on the above pictures and inspection results, we consider that the failure of enveloping heads of oxygen separator is mainly caused by the combination of intercrystalline corrosion, stress corrosion and electrochemical corrosion due to active ions, such as Cl and S2 , etc. These corrosions are closely related to the material, the processing of vessels, the electrolyte and working conditions. 5.1 Inter-Crystalline Corrosion. With the above inspections, it is illustrated that this corrosion was caused during welding enveloping heads. Since the material of stainless steel has been already treated by solid solution for delivery, the solid solubility of carbon in austenite will decrease with the drop of the
temperature. For example, at 1100C, the solid solubility of carbon in the material 1Cr18Ni9Ti is 0.2 percent, but it is only about 0.02 percent 1 at 500 700C. So the material of austenite stainless steel is over-saturated carbon after solid solution treatment. When the enveloping head is welded together with the cylinder body and the temperature of straight edge of the enveloping head is at a transition temperature of 450C 850C, the compound (FeCr) 23C6 will be formed with carbon segregated from austenite, and distributed on the crystal boundary 1 4. The concentration of Cr in compound (FeCr) 23C6 is much higher than that in normal austenite. The segregation of Cr results in the consumption of Cr adjacent to the crystal boundary. Furthermore, the consumed Cr can not be supplemented by diffusion from crystal particles because of slow diffusion speed. Therefore, the concentration of Cr adjacent to crystal boundary is lower than its deactivation limit
Fig. 3 The cracks extend along crystals boundary and there is transcrystallization in some area
Fig. 6 The cracks have branches at the tips and transcrystallization in some area
13 percent 5,6, and a Cr-poor area is formed. The potential near the crystal boundary is reduced for the state of deactivation is destroyed. Thus, a micro-couple cell is formed by the crystal particles with higher potential and crystal boundary, which is charac-
terized with a larger area of cathode and a smaller area of anode 5,6,9,10 that causes the corrosion in the crystal boundary area. 5.2 Stress Corrosion. The causes of stress corrosion are the following. Firstly, while the enveloping heads mold was pressed by cold-rolling, the internal structure of crystal was changed and the austenite was changed into deformed martensite, then malposition and slide happened at crystal boundary that initiated micro cracks. Secondly, the stress was caused during welding because of uneven heating. Finally, during operation, internal pressure was aroused with system pressure increasing. If all the above existed together, the stress corrosion would be initiated very easily in a vessel full of electrolyte. 5.3 Electrochemical Corrosion. It is well known that in alkaline solution, stainless steel material with Cr-Ni is easy to
Fig. 11 The components of the tips of cracks are composed of Cl, C, O and Fe, etc.
Fig. 12 The components of fracture are S, Cl, K and Ni, etc. Fig. 14 A huge corrosion cell is made up between hydrogen separator and oxygen separator, in which the oxygen separator
form a coating of oxide to prevent the electrochemical corrosion. If there is no Cl and S2 in the electrolyte, this kind of material could be used in alkaline solution at a temperature over 100C in accordance with related documents 3,8. However, the Cl is of highly active negative ions, of which ion radii are very small, and it can penetrate the oxidized coating easily and cause partial corrosion. With a continuous dissolution of the metal, Cl and S2 ions will move to maintain neutral electrically, thus Cl and S2 ions will be concentrated at the tips of cracks and on the crystal boundary, causing self-catalytically corrosion because of high concentrations of Cl and S2 ions 6 9, and this effect would result in a crack immediately. In such a process, the effect of oxygen could not be neglected. As the oxygen concentration is different in hydrogen separator and oxygen separator, a huge corrosion cell will be established, in which the oxygen separator is regarded as an anode, another as cathode, shown in Fig. 14. Furthermore, there are some disgurements in oxygen separator, a thick oxidized coating will not be formed and the corrosion is accelerated at the crystal boundary and cracks. Therefore, the enveloping head of oxygen separator will be in failure while the hydrogen separator does not. Meanwhile, attention should be paid to that: the Cl concentration in the electrolyte used in failed vessels is, in general, over 100 mg/l, and reaches to 500 mg/l in some samples. It is reported that the concentration of Cl should be controlled below 100 mg/l 2,4. However, studies have been performed on the same conditions in a high concentration of Cl and S2 of electrolyte. Among these sets of oxygen separators, cracks occur on any of them, but not on the others. By means of comparably research, the failure easily happens on the vessel with large stress and internal defect because of a large content of the deformed martensite generating highly magnetic while the vessel of no crack is found without any magnetic feature, so it shows that
this kind of vessel is processed with good solid solubility of austenite and little residual stress and the oxide coating is formed thickly and densely. Even if the contents of Cl and S2 are high, they can not penetrate the coating to cause corrosion.
Conclusion
Causes of the Failure on Oxygen Separator are as Fol-
6.1 lows.
i. During the process of manufacturing, annealing of the enveloping heads was not well and properly done to remove the stress and materials were not well treated by solid solution. All these causes resulted in defects such as micro cracks, malposition and/or slip, etc. ii. Because of the segregation of Cr on crystal boundary, a micro-couple cell would be formed and resulted in an intercrystalline corrosion. Nevertheless, the segregation was caused during welding, and the area inuenced by the heating was at a transition temperature (450C 850C). iii. In an electrolyte containing Cl and S2 ions and a high concentration of oxygen, the intercrystalline corrosion and the stress corrosion would occur rstly, which caused higher concentrations of Cl and S2 ions at the crystal boundary and the tip of cracks resulted in self-catalytically action. A large corrosion cell was formed from oxygen separator and hydrogen separator. All of those would speed up the corrosion. A conclusion is drawn that the failure of oxygen separator was caused by the combination of intercrystalline corrosion, stress corrosion and electrochemical corrosion. 6.2 Measures Should be Taken in Accordance With the Above Analysis. i. The chosen material should contain a lower content of carbon and a high content of molybdenum and titanium to reduce the segregation of carbon. ii. After modeled, a solid solution-treatment with tempering should be performed so as to remove the stress and keep carbon distributing well in material. iii. After the enveloping head and cylinder of separator were welded, the solid solution-treatment should be done once more and the temperature should be kept on 950C 1050C for two hours. iv. The concentration of Cl and S2 in electrolyte should be controlled under the normal condition. This could be met by choosing KOH of high quality with low concentration of Cl NOVEMBER 2004, Vol. 126 417
Fig. 13 The compositions of corrosion products are in two phases, the main compositions are Ca3 AlF SO4 OH 8 and Fe3 O4 .
ions. The concentration of Cl ions in pure water should be less than 2 mg/l, and that of Cl ions in the electrolyte should be less than 50 mg/l. v. During the vessel checkout, magnetic measurement should be made all over the area of enveloping head by a ferrite measuring instrument. Vessels of more than 1.5 percent magnetic phase should be retreated by solid solution.
Acknowledgment
The authors would like to acknowledge Huitang Shen, a professor, and many colleagues for their help in this work.
References
1 Zhang, Y., 1993, The Principle of Metal Corrosion, Aviation Industry Press, Beijing.
2 Huang, Y., 1989, The Principle of Metal Corrosion and Protection, Shanghai Jiao Tong University Press, Shanghai. 3 Sedriks, A. J., 1979, Corrosion of Stainless Steels, John Wiley, New York. 4 Zhang, Z., 1982, Partial Corrosion of Stainless Steels, Science Press, Beijing. 5 Lnoue, H., Koseki, T., and Ohkita, S., 2000, Formation Mechanism of Vermicular and Lacy Ferrite in Austenitic Stainless Steel Weld Metals, Sci. Technol. Weld. Joining, 56, pp. 385396. 6 Wei, B., 1996, Theory of Metal Corrosion and Application, Chemical Industry Press, Beijing. 7 Kotzmanidi, EM. Sarris, and Vassiliou, P., 1999, Effect of Heat Treatment in Reducing Plasma Environments on Chloride Ion Removal and Corrosion of Oxidized Steel Artifacts, Br. Corros. J., London, 344, pp. 285291. 8 Li, M. C., Zeng, C. L., and Cao, C. N., 2001, Electrochemical Corrosion Behavior of Type 316 Stainless Steel in Acid Media Containing Fluoride Ions, Br. Corros. J., London, 363, pp. 179183. 9 Cao, C., 1994, Electrochemistry Corrosion, Chemical Industry Press, Beijing. 10 Shiyingu, L., and Dekang, Zh., 1977, Stress Corrosion Cracking of Stainless Steels, Science Press, Beijing.