FINAL REPORT Part I DETERMINATION OF METALLURGICAL AND CORROSION FACTORS RESULTING IN (1) LOCALIZED GRAPHITIZATION ATTACK ON CAST IRON AND (2) BRITTLE FRACTURE OF MILD STEEL DISTRIBUTION PIPE By R. F. Hochman Metallurgy Department School of Chemical Engineering GEORGIA INSTITUTE OF TECHNOLOGY Atlanta, Georgia 30332 February, 1979
FINAL REPORT Part I I. ORGANIZATION OF FINAL REPORT This final report is submitted in two parts. This, the first part will cover the basic objectives, background and brittle fracture of mild steel and the second part will provide an indepth masters thesis study of the effect of stress on graphitization corrosion. The work on embrittlement of low carbon pipeline steel will continue in another graduate thesis study to start, in all propability this fall. II. OBJECTIVES The objective of the program's basic studies is to determine the metallurgical and mechanics relationships leading to the brittle fracture of mild steel, gas distribution piping, and to develop an understanding of the basic metallurgical characteristics, mechanical (stress) effects and environmental conditions leading to highly localized graphitization attack in centrifugally cast, cast iron. III. BACKGROUND A. General. Two distinct types of gas distribution line failures, which have received little attention, have been observed in actual service failures on several occasions by the project director of this program. The uniqueness of these failures require basic metallurgical and corrosion research to establish their exact mechanisms. These failures are the brittle cleavage fracture of ductile mild steel distribution lines under 10 to 30 pounds of pressure natural gas, and highly localized graphitzation corrosion, of centrifugally cast and sand cast, cast iron, the latter when originally laid, was advertised with a
minimum of 70 to 100 years corrosion life and now, although cast iron pipe are no longer permitted, those in service are suffering increasing failures annuallly. Since the initiation of this program more than three years ago five more new cast iron pipe failures were examined and four showed highly localized graphitization corrosion. It must be noted from reports and information from the Office of Pipe Line Safety (Department of Transportation) that corrosion is the major controllable factor in gas line failures and accounts for more than 5% of reported gas leaks and some of its most catastrophic accidents. B. Brittle Fracture of Mild Steel Materials in Gaseous Environments. Over the past several years, a number of distribution pipe fractures have been examined by the author. Typical of the fracture interface was the brittle cleavage of an otherwise mild steel. Figure 1 is a scanning electron micrograph of the brittle cleavage observed in mild steel pipe. Figure 1. Scanning electromicrograph CSEM) of the brittle fracture of surface of a mild steel distribution pipe. The author has examined, and verified, that all the cases in the attached list have exhibited brittle cleavage fracture of mild steel as the 2
the cause of failure. The typical mechanical properties of these materials are shown in Table 1. TABLE I Tensile Strength Yield Strength % Elongation 48,000 to 55,000 psi 32,000 to 37,000 psi 26% to 31% Failures have been found in natural gas constituents, for example in high hydrogen concentations it has generally been found that blistering occurs and that the hydrogen penetrated the steel to react with carbon and sometimes sulfur to cause blisters as a result of gas pockets. Rupture of the steel may also result. However in this common gas failure the metal fails in a ductile manner and must be ruled out for cleavage fracture. Stress corrosion, per se, must also be ruled out since in most steels the fracture is intergranular and not transgranular cleavage. However, in the past few years, it has been shown that mild steels and iron under stress at relatively low partial pressures, 15 pounds, have actually produced brittle failure. The work of J. P. Fidelle in France and others in this country in atomic energy research have produced brittle fractures in iron and a number of ductile materials including nickel and high nickel alloys (1). The test program and research studies of this program were geared to evaluate the embrittlement of these steels by internal gas and outside corrosion effects at the threaded joints. (1) L'Hydrogene Dans Les Metaux, J. P. Fidelle, Colloque-Valdue, France 1967. 3
IV. EXPERIMENTAL RESULTS OF In addition to these normal mechanical tests, a series of subsidized impact tests were made to evaluate transition temperature characteristics. Tests at temperatures to freezing did not show a brittle fracture. Only -40 C tests indicated evidence of brittle fracture. Because of the uniqueness of this type of failure, it was necessary to examine the conditions which could bring about such unique metallurgical fractures. Four factors were evident: 1. The pipe in each case contained stagnant gas or was at an unused deadend of a line. 2. Some type of mild stress, "in two cases the result of its own weight," had been applied to the pipe. It appeared that although stresses could not be determined that they were well below normal yield stress for the materials. 3. The fracture occurred at an exposed thread in a coupling or a T area. This could be the result of a stress concentration increase in these areas, but certainly not enough to produce the type of fracture observed. 4. Steels are basically ductile in the.05 to.08% carbon range. It was necessary to establish a series of tests and a test program to determine what was causing this type of fracture, first because of the high probability of its increasing occurrance in the future, and second, its overall importance to steel gas line systems. Several individual cantilever beam tests of 3/4" ID pipe with natural gas, hydrogen and natural gas plus water vapor environments were tested at loads from 50% to 80% of the yield stress at the threaded root. Time of testing, difficulty in individual set ups, and variation in tests results required that a more consistant, reproducable system, providing statisical data with a 4
greater number of tests was necessary. Therefore, the test system shown in Figure 2 was developed. This entails the use of a series of cantilever beams. Figure 2. The typical cantilever beam test system for combination stress and internal environment testing. With this system multiple tests including purging, evacuating, and gas changing with the desired gas composition were easily conducted. To date, tests have been made at 15 psi with (a) hydrogen, (b) hydrogen plus 1% moisture, (c) natural gas, (d) natural gas plus 1% moisture, (e) low pressure natural gas and (c) and (d) with moisture at the outer surface at the thread roots. This was accomplished by moisturizing a cloth wrapped around the thread roots. The results of the test performed are given in the following Table II. 5
TABLE II Environment Tensile % of Yield Stress at thread No. Fracture Test to 50% 900 ) Tests Failure Ductile * * t I (a) Hydrogen 8-24 hrs 6 Partial Brittle,. w 7:1 * * (b) H2 moisture 4..) a) 4-1 m 4.) 2-4 hrs 4 Partial Brittle I I (c) Natural gas a)..-1 8-24 hrs 6 Partial Brittle 4-) * (d) N.G. + 1% moisture H 2-6 hrs * 8 * Partial Brittle m0 N.G. + 1% H 2 C.) ± 2-5 hrs cn m 8 Partial Brittle t4-1 r-i * * (e) Low P.N.G. I t 2-6 hrs 8 Partial Brittle (f) (c) + corrosive environment at the threads 11 (g) (d) + corrosive environment at the threads z o v * 31/2-6 hrs * 8 Partial Brittle 21/2-51/2 hrs * 8 Partial Brittle (1) Rechecks showed test loads for the accelerated tests ran from 90 to 96% of the actual yield. Checks were made to insure no yielding during the test. These studies indicate that hydrogen in stagnated gas filled pipes may be a major factor in the brittle fractures found in a number of mild steel distribution lines. However, more cleavage has been found in natural gas plus moisture, than in pure hydrogen or pure hydrogen plus moisture. Thus effects of gaseous components as well as metallurgical effects and stress intensity also play an important role in these failures. In each fracture case the temperature has been sufficiently high (70 to 85 F) to rule out nil ductility fracture. A very unique problem resulted in the difference between galvanized and black pipe. The black pipe has a larger OD, but also a larger ID with the resultant thread cut leaving a wall thickness of only.052 inch at the thread for black pipe versus.068 for galvanized pipe. However, in the case of galvanized pipe, it appears that the corroding of the galvanized layer may produce excess hydrogen at the bare steel areas. 6
The results have shown that brittle fracture can be duplicated in stagnant gases under loads below the yield point. This type of brittle cleavage was studied under accelerated tests. Longer term tests to determine what the lifetime might be under 50% yield stress at the thread concentration area did not result in fracture in a convenient time. It appears that not only moisture, but the possibility of small concentrations of sulfur bearing compounds may also have increased embrittlement tendencies. These factors are planned for future investigation. Figure 3. A SEM fractograph of a specimen tested as per (g) in Table II. Essentially no literature or information is available on brittle or cleavage fractures of mild steels in natural gas except as, observed in the studi\es in this program and some NBS hydrogen work (2). Some mild steels show small amounts of cleavage fracture if an inherent brittleness exists which (2) Private communication from J. Kruger of the NBS. 7
can be attributed to dissolved hydrogen, nitrogen, and possibly oxygen, but again this brittleness is not well documented and has not been related to the effect of stagnated natural gas. Further data, gathered in this study, will complement this review. V. CONCLUSIONS TO DATE ON THE BRITTLE FRACTURE OF MILD STEEL DISTRIBUTION PIPE A broad range of metallurgical variables and the effect of impurity elements on the embrittlement phenomena must still be determined. However, this limited effort of approximately $20,000 has pinpointed several major features of the embrittlement phenomena: A. A unique difference was found between galvanized and black pipe. The wall thickness of the black pipe will have a 0.052" minimum wall at the root of thread cuts versus a wall thickness of 0.068" remaining in the same area for the galvanized pipe. B. Tests in one atmosphere hydrogen plus 1% moisture resulted in reduced time to fracture and an increase in the amount of brittle fracture observed. A similar effect is found when moisture and hydrogen are added to 1 atm. of natural gas. C. Low pressure natural gas (1 atm.) plus 1% moisture plus an exterior corrodant at the threaded area produces a fracture with more brittle area than in other tests. D. Corrosion on the outer surface of galvanized pipe may produce hydrogen on the bare steel (cathodic area) in threaded areas increasing brittle fracture characteristics. Respectfully submitted, r - - /Robert F. Hochman Project Director 8