Materials Chemistry
Creep-Fatigue Damage for Boiler Header Stub Mock-Up Specimen of 47Ni–23Cr–23Fe–7W Alloy
2020 Volume 61 Issue 6 Pages 1109-1114
Details
2020 Volume 61 Issue 6 Pages 1109-1114
Header stub welds for thermal power plants undergo creep-fatigue damage due to thermal expansion and contraction, which are caused by cyclic startup and shutdown of the plant. In this study, creep-fatigue tests were conducted using header stub mock-up specimens of 47Ni–23Cr–23Fe–7W alloy to investigate the creep-fatigue damage process. It was discovered that the cracks were initiated from the outer surface of the tube near the bond line, and they propagated toward the inner side in both fatigue and creep-fatigue tests. Transgranular cracks were observed in fatigue tests, whereas cracks were found to be progressed along the grain boundary in creep-fatigue tests. The failure mechanism of 47Ni–23Cr–23Fe–7W alloy header stub mock-up specimens was composed of two steps. At first, cracks progressed on the surface near the bond line, and then progressed toward the inner surface. Crack initiation and propagation behaviors of the 47Ni–23Cr–23Fe–7W alloy header stub mock-up specimen were the same as those of the 2.25Cr–1Mo steel header mock-up specimen.
This Paper was Originally Published in Japanese in J. Soc. Mater. Sci., Japan 68 (2019) 136–141.
In the thermal power generation sector, efforts are being made to increase steam temperature with the aim of improving power generation efficiency and reducing CO2 emissions. For the A-USC (Advanced Ultra Super Critical) boiler currently under technology development, studies are being conducted on the use of Ni-based alloys instead of conventional steel materials for some components such as the main steam piping from the standpoints of creep strength and high temperature corrosion because steam is superheated to more than 700°C in this boiler. Toward the application of Ni-based alloys, element tests have been conducted whereby welding and bending conditions were defined and then mock-up specimens were fabricated based on such conditions.1,2) Moreover, the element tests were followed by actual boiler tests using steam at temperatures of over 700°C, whereby it was proved that the boiler was able to operate without problems.3)
The next issue to be addressed for the A-USC boiler is the establishment of maintenance technology toward its practical use. Especially for parts that have seen damage in existing boilers, we believe it is necessary to ascertain failure mechanisms and develop assessment techniques for the application of Ni-based alloys to them. Damage that has occurred in existing boilers includes creep damage of stub piping, headers, etc.;4,5) creep damage of piping welds;6,7) stress corrosion cracking of stainless steel circumferential joints;8) and reheat cracking of such joints.9) In addition, reheat cracking of girth welds of Ni alloy piping has been experienced in actual scale testing conducted in Europe,10) and this fact suggests the importance of weld assessments. Hence, in order to ascertain the mechanisms of such damage, the following activities have been carried out for the alloy 47Ni–23Cr–23Fe–7W (HR6W), which is a prospective material for the piping of A-USC boilers: creep assessments of headers and circumferential joints;11) unraveling of damage mechanisms through a large-scale creep test;12) and a study of reheat cracking and conditions for stress-relief annealing.13) However, no studies have been conducted on stub welds heretofore. For ferritic heat resisting steels, which are used in existing boilers, a study on the mechanism of damage resulting from creep fatigue was previously conducted, succeeding in the reproduction of damage morphology.14) Thus, we believe this technique can be used to unravel the process through which the stub welds of the A-USC boiler become damaged.
In this research, we therefore aimed to unravel the mechanism of creep-fatigue damage of header stubs made of Ni-based alloy. With this aim, we fabricated stub mock-up specimens of HR6W and then conducted creep-fatigue tests on them at a temperature of 750°C, to which they would be exposed in an actual A-USC boiler. Each creep-fatigue test consisted of rupture testing and creep-fatigue interrupted testing. First, the rupture testing was conducted to check whether the morphology of the damage accurately represented the damage that might be suffered by an actual A-USC boiler. The creep-fatigue interrupted testing was then conducted to study crack propagation behavior.
Stub mock-up specimens with the shape shown in Fig. 1 were fabricated by welding a small-diameter tube of HR6W (45 mm in diameter, 9.3 mm in thickness) to a plate of HR6W (50 mm in thickness). Both the tube and the plate were subjected to solution heat treatment and then TIG-welded to each other. Alloy 617-based flux was used as the welding consumable. The tube-and-plate assembly was then subjected to post weld heat treatment at 900°C for three hours13) to prepare for testing. For the purpose of comparison, smooth round-bar specimens were taken from a large-diameter tube, which were then subjected to high-temperature fatigue testing. The configuration of these specimens was as follows: 10 mm in diameter in the parallel section; 30 mm in length. A test temperature of 750°C and five strain ranges (Δε = 0.5%, 0.7%, 1.0%, 1.2%, 1.5%) were specified as the test conditions. Table 1 shows the chemical composition of the large-diameter specimens and the small-diameter specimens.
The specimen simulated for boiler header stub.
Figure 2 shows a stub mock-up specimen in the testing equipment. Uniaxial fatigue testing equipment was used. The plate member of the stub mock-up specimen was fastened to the U-shaped jig attached to the lower piston. In addition, the distal tip of the small-diameter tube member was screwed into the threaded hole located in the upper fixture having a diameter equal to that of the small-diameter tube and then fastened. Due to the presence of about 0.1 mm of play between the threaded hole and the small-diameter tube, the amount of play was added to the displacement predetermined for testing. A bending load was then applied to the weld bond line on the small-diameter tube by moving the lower jig up and down. Displacement control was adopted as the test control method out of consideration for the damage modes of the actual boiler.
Test equipment.
Meanwhile, the stub mock-up specimens were complicated in configuration, and the amount of strain in the parts subjected to the bending load was expected to vary depending on the displacement. In addition, the amount of strain introduced varies from location to location even if the locations are very close to each other. Given this, we conducted a study of local strain introduced into a weld bond line in order to determine the displacement used for these tests. Although the test temperature for the present test was 750°C, we decided to take measurements of local strain at room temperature and use them as alternative strain in this study due to the inability to measure strain at high temperatures. With a strain gauge placed across a weld bond line, as shown in Fig. 3, the jig was moved within a 0–2.0 mm range so as to derive the relationship between the displacement and the axial local strain of the weld bond line. In order to minimize the impact of this operation on the straight tube section, a strain gauge with a gauge length of 2.0 mm was used to take measurements.
Schematic image of measurement place of local strain.
Figure 4 shows the displacement-local strain relationship at room temperature. When the displacement was within a 0–2.0 mm range, strain had a linear relationship with displacement. This relationship can be expressed by the following equation:
| \begin{equation} \varepsilon = 0.2823 \times \mathrm{S} - 0.0203 \end{equation} | (1) |
Relationship between stroke and local strain.
Table 2 shows the test conditions determined for the present testing. A test temperature of 750°C was set for both the fatigue testing and creep-fatigue testing. Each specimen was heated by using the high-frequency induction heating method while the temperature was controlled to keep the temperature of the areas on both sides of the bond line on the small-diameter tube (i.e., the small-diameter tube side and welded metal side) within a 10 mm range of the bond line within 750°C ± 5°C. First, high-temperature fatigue testing was conducted on stub mock-up specimens and the number of cycles to failure was compared with the result of a uniaxial test. This fatigue testing was followed by creep-fatigue testing, wherein a displacement equal to that for the fatigue testing was used. In addition, interrupted testing was conducted at life ratios ranging from 17% to 61% using the same specimens in order to ascertain crack propagation behavior on their surfaces. After the testing had been interrupted, penetrant testing (PT) was conducted to ascertain crack locations and measurements were taken of the lengths of surface cracks at the respective life ratios. Moreover, in order to ascertain the depths of cracks in the thickness direction, four specimens designed to fracture at a life ratio of 38% to 100% were fabricated for interrupted testing and a fractographic study was conducted on them. Measurements were performed on each fracture surface obtained, in order to determine the maximum length of the vertical section of the crack propagation domain.
In order to determine the test conditions for these tests, stub simulation fatigue testing was conducted first as preliminary testing. Figure 5 shows the curve of the number of cycles to failure versus load at a test temperature of 750°C and at a displacement of 3.0 mm. Incidentally, due to the absence of a clear definition of the number of cycles to failure for stub-shaped specimens, “the number of cycles at the moment the tensile load decreases to 3/4 its peak value” was defined as the number of cycles to failure in line with the definition of Ito, et al.14) In the stub simulation fatigue testing, a constant load was maintained within a certain range of cycles and then suddenly dropped, as seen in a uniaxial fatigue test of a general work hardened material. In addition, the calculated number of cycles to failure was 980.
Cycle-load curve for fatigue test.
Next, in order to make sure that the test conditions for the stub simulation testing were within the domain of low cycle fatigue testing, round bar specimens were taken from large-diameter tubes of HR6W and fatigue testing was conducted on them at 750°C. Figure 6 shows the obtained results. The number of cycles to failure for the round bar specimens met the standard of the Society of Materials Science, Japan.15) At a strain range of 0.83%, the number of cycles to failure was about 750. Thus, the test conditions for the stub simulation fatigue testing were thought to be within the scope of low cycle fatigue testing. Given this, a displacement of 3.0 mm was set as a test condition for creep-fatigue testing in the present research.
Relationship between the number of cycles to failure and strain range.
Figure 7 shows the curve of the load versus the number of cycles to failure after creep-fatigue testing, wherein displacement continued to be applied for 60 minutes. The curve indicates that a constant load of approximately 17 kN was maintained within a certain range of cycles and then suddenly dropped. This result matches the trend of the load-number of cycles relationship as seen in ordinary creep-fatigue testing. In addition, the calculated number of cycles to failure was 184.
Cycle-load curve for creep-fatigue test.
Figure 8 shows the appearance of specimens that underwent stub simulation fatigue testing and the appearance of specimens that underwent creep-fatigue testing. Both specimen groups showed that a crack formed near the bond line of the small-diameter tube and propagated along the bond line. In these tests, no cracks were found in areas distant from the bond line. Next, Fig. 9 shows the results of cross-sectional observation of crack propagation domains. Whereas the cracks that formed in the fatigue test specimens propagated transgranularly, the cracks in the creep-fatigue test specimens propagated intergranularly. It seems that creep contributed to the difference in crack propagation route between the specimen groups. A uniaxial creep-fatigue test of HR6W demonstrated that the creep propagation route varied depending on the retention time16) and this result is consistent with the result of the present testing.
Appearances of fractured specimens (a): After fatigue test (b): Larger image of (a) (c): After creep-fatigue test (d): Larger image of (c).
Optical micrographs of cross section images for fractured specimens (a): After fatigue test (b)–(c): Larger image of (a) (d): After creep-fatigue test (e)–(f): Larger image of (d).
In order to understand crack propagation behavior on surfaces, creep-fatigue interrupted testing was conducted on stub mock-up specimens. The same specimens were consistently used for testing. After the testing had been interrupted at each life ratio, PT was conducted on the specimens. Figure 10 shows typical examples of the PT results. In this connection, the number of cycles to failure for this interrupted testing was defined as the number of cycles at a life ratio of 100%.
Appearances of PT inspection under creep-fatigue interrupted tests (a): Life ratio: 17% (b) Life ratio: 24% (c) Life ratio: 41% (d) Life ratio: 61%.
At the interruption at a life ratio of 17%, as shown in (a), PT indications attributable to cracks were not observed. Meanwhile, at the interruption at a life ratio of 24%, as shown in (b), microcracks were observed near weld bond lines. Thereafter, with an increase in life ratio, crack length increased in the circumferential direction, as shown in (c) and (d).
Figure 11 shows the results of measurements of aggregated crack lengths at the respective life ratios. Cracks formed at a life ratio of around 20% and then rapidly propagated along the circumference until the life ratio reached around 40%. Thereafter, crack lengths were nearly constant.
Relationship between crack length on the surface and life ratio under creep-fatigue test.
Next, we conducted a study of crack propagation behavior in the thickness direction from the formation of cracks until the occurrence of failure. For the purpose of creep-fatigue interrupted testing wherein interruptions occurred at multiple life ratios, we fabricated specimens for each life ratio. In order to observe fracture surfaces at the times of interruptions, we conducted room temperature fatigue testing whereby the fracture surfaces were opened. Figure 12 provides photographs showing external views of the obtained fracture surfaces, as well as those showing enlarged views of the oxidized regions in the fracture surfaces.
Appearances of fracture surface after creep-fatigue interrupted tests (a): Life ratio: 38% (b) Life ratio: 43% (c) Life ratio: 61% (d) Life ratio: 100%.
The opened fracture surfaces, including that of the specimen for interruption at a life ratio of 38% as shown in (a), indicated the formation of cracks during the test and the presence of oxidized regions localized on the outer surfaces. As shown in (b) through (d), the oxidized regions extended in the thickness direction with an increase in life ratio. These oxidized regions matched the crack propagation domains observed in the creep-fatigue testing. We therefore took measurements of the maximum length in the thickness direction of each oxidized region, thereby deriving relationships between life ratios and depths of crack propagation in the thickness direction. Figure 13 shows the relationships between life ratios and crack depths in the thickness direction. Starting from a life ratio of around 40%, cracks in the thickness direction monotonically propagated. The number of cycles at a life ratio of 40%, at which such crack propagation in the thickness direction was observed, closely matched the number of cycles at the moment the tensile load decreased to 3/4 its peak value, as shown in Fig. 7. This fact, along with the surface crack propagation behavior shown in Fig. 11, suggests that cracks formed by the creep-fatigue testing started to propagate along the circumference first at a life ratio of around 20% and then propagated in the thickness direction starting from a life ratio of 40%, and finally fractured. Given this, it is important during the maintenance of a header stub made of HR6W to check for crack initiation on the outer surface near the bond line.
Variation of crack depth with life ratio under creep-fatigue test.
Creep-fatigue crack propagation behavior studied with stub mock-up specimens of 2.25Cr–1Mo steel has been reported by Ito, et al.13) Their report states that in the stub mock-up specimens of 2.25Cr–1Mo steel, cracks formed on the outer surfaces of the tubes and then propagated along the circumference first, then propagated in the thickness direction, and finally fractured. The crack propagation behavior derived from the present testing was analogous to that showed by 2.25Cr–1Mo steel. Thus, a comparison was made between stub mock-up specimens of HR6W and those of 2.25Cr–1Mo steel with regard to the relationships between life ratios and depths of cracks in the thickness direction. Figure 14 shows the relationships between life ratios and depths of cracks in the thickness direction for both HR6W and 2.25Cr–1Mo steel. In order to enable the data from the present testing to be directly compared with the data collected by Ito, et al., values equal to a crack depth divided by the thickness of the small-diameter tube were assigned to the vertical axis. In addition, the number of cycles at the moment the crack depth reached 25% of the thickness was redefined as a life ratio of 100% and was assigned to the horizontal axis. The crack depths relative to life ratios for the stub mock-up specimens of HR6W monotonically increased starting from a life ratio of 60%. This trend closely matched the results of the testing of stub mock-up specimens of 2.25Cr–1Mo steel. Although the reason for the consistency in life-crack depth relationship between different materials is unknown, one possible reason for the consistency in crack formation and propagation behavior between these materials is the contribution of stress concentration due to the stub configuration.
Change in depth of crack growth of HR6W and 2.25Cr–1Mo steel for creep-fatigue test.
With the objective of understanding the creep-fatigue properties and damage mechanism of stub mock-up specimens of HR6W, we conducted fatigue testing and creep-fatigue testing. The following findings were obtained.
