Development of a 1,000°C class creep testing machine for ultraminiature specimens and feasibility verification

2.1 Overview of the testing machine

Figure 1(a)–(c) shows an overview schematic illustration of the front and side views of the developed creep testing machine for using ultraminiature specimens, respectively. The full size of the creep testing machine is 1,200 mm in height, 600 mm in width, and 600 mm in length. The developed machine is an electric furnace heating creep testing machine. During the creep test, creep strain was recorded via two linear variable differential transformer (LVDT) set at the lower middle portion of the machine. To avoid overheating of other components during the test, a heat shield plate made of Al alloy was mounted on the machine, as shown in Figure 1(b). The stress can be loaded using weight, and the lab jack is set beneath the weight cushion shock to prevent the weight from breaking the testing machine when the tested specimen fractures. The material used for the components inside the electric furnace was AISI 310S stainless steel due to its excellent high-temperature resistance. The other components were made of AISI 304 stainless steel. To avoid overheating of the shaft during the creep test, a water cooling system was introduced. Data can be collected using a data logger.

Figure 1

(a) Overview of the developed creep testing machine for ultraminiature specimens. (b) Schematic illustration of the front view and (c) Side view.

The specifications of the developed creep testing machine for using ultraminiature specimens are listed in Table 1. The maximum testing temperature and applied loading are 1,200°C and 200 N, respectively. The loading mode and displacement measuring approach are described in the previous section. The testing environment is air.

2.2 Initial loading reduction structure

Based on the JIS creep testing standard, the initial loading that can be applied to a specimen during a temperature increase is less than 10% of the objective loading. Since this testing machine is a direct loading type, the specimen is loaded by the deadweight of the shaft, and the weight holder increases the initial load during temperature increasing, which leads to increased initial loading. As a result, the lower limit of the test loading that satisfies the JIS standard increased, making it impossible to perform tests at low loading. Therefore, as shown in Figure 2(a), a structure to reduce the initial loading during temperature increasing using an auto-lab jack and a compression coil spring was developed. For the structure, a compression coil spring contacts with a displacement measurement bar that connects the shaft by compressing the spring, and the reduction of the initial loading by the generated elastic force can be achieved (Figure 2(b)). The amount of spring compression can be maintained at a constant elastic force by adjusting the height of the auto-lab jack, and the initial loading is kept constant during temperature-increasing period (Figure 2(b)). After completing temperature increasing and holding period, the compressed spring will be released, and the set specimen will be loaded (Figure 2(c)).

Figure 2

Schematic illustration of the initial loading reduction structure for the specimen using the auto-lab jack and the springs. (a) Pre-heating stage; (b) Temperature increasing stage; and (c) Loading stage.

2.3 Alignment structure

Because the testing machine is of a uniaxial tensile loading type, the loading and longitudinal directions of the specimen must coincide; otherwise, a large error will be caused in the fracture time. Therefore, alignment of the axes of the test piece, shaft, chuck, weight receiver, etc., is essential. To achieve the above goal, the rod, shaft, chuck, etc., were connected with pins, and the shaft and weight holder were connected using wire, lead balls, etc.

The pin connection and its function are shown in Figure 3(a) and (b). The pin acts as a free joint, and the angle of the connected shaft and chuck can be changed freely. As shown in Figure 3(b), if the rod, shaft, and chuck are integrated rather than pin-connected, the rod axis is not in the vertical direction. This, in turn, suggests that loadings will be applied vertically while the shear force will be applied to the cross section of the gauge portion of the specimen. In contrast, in the case of a pin connection, each axis coincides when loadings are applied identically, and alignment is achieved, clearly confirming that shear stresses cannot be generated on the cross-section of the gauge portion of the specimen, as only the tensile force is applied along the vertical direction.

Figure 3

Schematic illustration of a pin connection: (a) Side and front view, and (b) Function for avoiding shear loading.

The conventional connection between the shaft and the weight holder was to hang the weight holder by fastening wires on both sides of the shaft, as shown on the left side of Figure 4. However, with the same structure, the weight holder tends to shift with respect to the axis of the shaft, indicating that alignment is difficult to achieve. Therefore, connecting the shaft and weight holder via lead balls and oval sleeves using wire was proposed, as shown on the right side of Figure 4, by which alignment is easy to achieve. When loading is applied, the lead ball slides into the tapered shape, automatically aligning the axes. Furthermore, the used wires were stainless steel with a diameter of Φ0.72 mm. The used wires need to be flexible so that axes alignment can be achieved when loading is applied. Previously, the use of stranded wires was considered because it has both flexibility and strength; however, stranded wires would unravel and cause the weight to rotate. Therefore, as shown in Figure A1, weights were sandwiched between the strings, and when the weights rotated, the tensile loading along the string prevented further rotation, but twisting deformation occurred in the specimen. Therefore, to prevent weight rotation, a single stainless steel wire with a diameter of Φ0.72 mm was used.

Figure 4

The connection of the shaft and weight plate with alignment.

Figure 5

Shape and dimensions of (a) Specimen A and (b) Specimen B. Illustration of alignment for (c) Specimen A and (d) Specimen B.

2.4 Specimen design

Two types of specimens were designed in the present study, and their shapes and dimensions are displayed in Figure 5(a) and (b). For specimen A, the total length, gauge length, and thickness are 14, 7, and 2 mm, respectively. In contrast, the total length, gauge length, and thickness of specimen B, which was first proposed in the study of Takahashi et al. [24], are 10, 5, and 0.5 mm, respectively. Moreover, the widths of the gauge sections for the two specimens are 1 and 2 mm, and the shoulder portions of specimens A and B were designed to be T- and tapered-shaped, respectively. It should be noted that one of the potential disadvantages of specimen A is that it is difficult to align the axis of the specimen with the axis of the jigs during the setting process. In contrast, this issue can be avoided in specimen B. Nonetheless, it cannot be concluded which type is better prior to conducting FEM analysis.

Figure 6

Established FEM analysis models for specimens (a) A and (b) B.

For the ultraminiature creep test, it is impossible to mount an extensometer or a similar device directly to the gauge portion of the specimen due to size limitations. Therefore, as described in Section 2.1, the creep strain until fracture of an ultraminiature creep specimen can be obtained by measuring the displacement of the shaft connected to the lower side of the specimen using LVDT. However, if the shoulder of the specimen undergoes creep deformation, the obtained creep strain is bound to include both shoulder and gauge deformations, thereby resulting in incorrect creep strain measurements. Therefore, as a test to verify the designed specimens, FEM analysis for creep was conducted by Marc Mentat (2023) to determine which specimen has an effective shape with less deformation in the shoulder portion.

Figure 6 shows the analytical model and boundary conditions for the two designed specimens. The employed material properties are listed in Table 2. The employed material properties were obtained from Norton’s law of another material. By fitting the graph, where abscissa and ordinate indicate applied stress and the minimum creep strain rate, respectively, the used Norton’s law can be acquired. The model of the plane stress element (4 nodes) for specimen A has 1,713 nodes and 1,600 elements, while 4,046 nodes and 3,860 elements are included for specimen B. To load a 180 MPa tensile stress to the gauge sections of the two specimens, edge loadings were used with directions set perpendicular to the shoulder portions.

Figure 7

Mapping of FEM creep analysis for specimens (a) A and (b) B.

Figure 7 shows the deformation mapping along the X direction of the gauge and shoulder portions of the two specimens after 300 h of FEM creep analysis. On the basis of Figure 7, it can be observed that the deformation of the gauge portion close to the R portion for specimen A is almost twice that of specimen B at the identical location. It demonstrates that the risk of fracture at the R portion for the former is higher than that for the latter. On the other hand, the deformations of the shoulder portion are extremely different between the two specimens. Deformations along the Y direction (arrows in Figure 7) from the central axis (red dash-dotted line in Figure 7) of specimens are displayed in Figure 8. For specimen A, the deformation of the shoulder portion was approximately 17 times higher than that of the gauge portion. At a location along the Y coordinate of approximately 2 mm, the deformation of the shoulder portion increased drastically, indicating that a large error will be induced when measuring creep strain using LVDT. In contrast, the deformations of both shoulder and gauge portions for specimen B were almost identical, and their values were also significantly low. Thus, on the basis of the above analysis, specimen B was considered to be a more effective specimen. In summary, considering the measurement of creep strain, including large deformation (displacement) for specimen A that leads to large data error, specimen B has been selected for the developed creep testing machine.

Figure 8

Deformations of the gauge and shoulder portions for specimens A and B.

Figure 9

Image of the thermocouple setting location for temperature measurement.

3.1 Temperature validation

The temperature distribution in the electric furnace is uneven during creep tests. Therefore, to investigate the temperature distribution behaviors at the objective gauge portion, and jig values, temperature-increasing testing was carried out. Figure 9 shows the mounting location of the two thermocouples, where one is mounted at the center of the specimen, and the other is mounted slightly below the lower pin. The objective temperature was set to 700°C and 1000°C, which is identical to that used for conducting the size effect testing.

Figure 10

Temperature validation results at (a) 700°C and (c) 1,000°C. (b) and (d) Enlarged graph of the area enclosed by a rectangle in (a) and (c).

The temperature validation results of the objective temperature of 700°C are shown in Figure 10(a) and (b), where the black, red, and blue symbols indicate the temperature of the gauge portion of the specimen, jig, and set values, respectively. After approximately 1 h, the temperature reached the set value. It can be observed that the temperature-increasing ratios of the gauge and jig portions were identical to those of the set values. Moreover, the magnified graph of the area marked with a green rectangle in Figure 10(a) is shown in Figure 10(b). When the gauge central portion of the specimen reached 700°C, the set temperature was approximately 690°C. In contrast, the temperature of the jig portion was close to 693°C. Using the temperature-increasing test, the difference between the set and gauge temperatures was confirmed. Therefore, for the following tests, the objective temperature was set at 690°C, and the temperature of the gauge portion was also measured for each test. Moreover, temperature validation results at 1,000°C are displayed in Figure 10(c) and (d), in which black and blue plots indicate the temperature of the gauge portion of the specimen and objective values. Similarly, after approximately 1.5 h, the objective setting temperature of 996°C was reached. Meanwhile, the gauge portion of the specimen was heated to 1001.9°C. Although at holding time close to 2.5 h, the temperature of the gauge portion of the specimen was slightly decreased to 999.6°C for a short time and increased again (Figure 10(d)), and taking into account there is a gap between the set and measured temperatures of the gauge portion, it could be considered that carrying out creep tests at 1,000°C is applicable for the developed machine.

Figure 11

Comparison of the results obtained via bulk and ultraminiature specimens: (a) Creep curves and (b) Creep strain rate.

3.2 Size effect test

A size effect exists in the mechanical testing field, which may lead to significant differences in the results obtained by bulk and miniature specimens or specimens of different sizes [25,26]. Thus, to investigate whether there is a size effect between the designed and bulk specimens, validation tests were performed with applied stresses of 150 and 120 MPa at 700°C in air. A bulk specimen with a gauge length of 30 mm and a diameter of 6 mm was used. The material used in this process was AISI 304 austenitic stainless steel, which has excellent mechanical properties and high-temperature resistance [27,28]. In the current study, to avoid the influence of different material batches on test results, both bulk and ultraminiature specimens were manufactured from identical batches.

A comparison of the results is shown in Figure 11. In Figure 11(a), the solid and hollow symbols denote the data tested using ultraminiature and bulk specimens, respectively. For tests with applied stress of 150 MPa, the creep fracture time of the ultraminiature specimen was slightly longer than that of the bulk specimen.

Figure 12

Comparison of results obtained via bulk and ultraminiature specimen: (a) Creep fracture curve, (b) Norton law, and (c) Monkman–Grant law.

Similarly, at an applied stress of 120 MPa, the fracture time obtained from the ultraminiature specimen was slightly shorter than that of the bulk specimen. Moreover, there was no significant difference in the creep fracture strain between the two types of specimens. Thus, it can be confirmed that creep strain measurements using LVDT are applicable to the designed creep testing machine. The variation of creep strain rate, which can be obtained by differentiating creep strain with loading time, is depicted in Figure 11(b). Similarly, only a slight difference was observed between the two specimens. Considering the nature of the scattered data in the creep test, it could be reputed that the creep curves and creep strain rate for the two specimens tested under identical stress levels agree with each other.

To further analyze whether the data obtained were influenced by the size effect, a comparison of the results obtained via bulk and ultraminiature specimens for creep fracture curve, their Norton relationship, and Monkman–Grant relationship is displayed in Figure 12(a)–(c), respectively. In Figure 12(a), as previously described, the creep fracture time for the two specimens tested under the same applied stress can be considered identical. The Norton relationship [29], which can be used to determine the creep deformation mechanism, is shown in Figure 12(b), where the abscissa and the ordinate denote the applied stress and minimum creep strain rate, which can be obtained from Figure 11(b), respectively. The dashed line is the fitting curve of the results tested with a bulk specimen. It can be observed that the ultraminiature tested results fall into the fitting curve, i.e., the Norton relationship for the two specimens matches with each other. In other words, the creep deformation mechanism is identical. Similarly, in the Monkman–Grant relationship [30] (Figure 12(c)), the abscissa and ordinate indicate the minimum creep strain rate and creep fracture time, respectively. The function of the Monkman–Grant relationship is to deduce the creep fracture time when gaining a minimum creep strain rate at a given applied stress level. Similar to Figure 12(b), the dashed line in Figure 12(c) is also the fitting curve of the results acquired via bulk specimens. Data plots of ultraminiature specimens agree with those obtained via bulk specimens. Fracture surface observations were performed after creep tests. Figure 13(a)–(d) represents typical fracture surfaces tested under 150 and 120 MPa. A reduction in the areas for the two fracture surfaces can be observed (Figure 13(a) and (c)). These high magnification images (Figure 13(b) and (d)) demonstrate the formation of dimples, i.e., fracture surfaces behaving with ductile fracture characteristics, which is consistent with the reported results of identical materials tested using bulk specimens. From the above discussion, it could be considered that the size effect on the creep fracture time for the investigated AISI 304 stainless steel has not been observed.

Figure 13

Fracture surface observation under 150 MPa: (a) Overview, (b) High magnification, and 120 MPa; (c) Overview and (d) High magnification.

Figure 14

EDS analysis results of a specimen tested under 120 MPa: (a) Overview of the profile, and (b) and (c) Magnified images marked in (a).

It is also well known that for small-sized specimens, the effective loading-bearing area reduction resulting from oxidation layer growth during creep testing is frequently observed for various materials. This may lead to the development of a gap in the fracture time between small-sized and bulk specimens [31,32]. To survey the oxidation behaviors of the ultraminiature specimen, the oxidation layer of the fractured specimen was observed. Because the oxidation layer grows with increasing heating time, the specimen tested under an applied stress of 120 MPa was selected due to its longer fracture time. Figure 14 shows the results of the energy-dispersive X-ray spectroscopy (EDS) analysis. In Figure 14(a), area 1 is far away from the fracture surface, and the corresponding EDS maps are shown on the right side of Figure 14(b). It can be observed that the thickness of the Cr-rich oxidation layer was approximately 1 μm after exposure for 288 h at 700°C in air. In contrast, a long crack was formed in area 2 because this is a closed fracture surface, and the corresponding EDS map is also shown on the right side of Figure 14(c). The thickness of the Cr-rich oxidation layer in the area was approximately 30 μm, which is induced by oxidation layer cracks during creep deformation. Subsequently, oxygen is allowed to contact the metal surface through these cracks to accelerate the formation of the oxidation layer. This phenomenon was also reported in creep tests using bulk specimens [33]. However, by the above-mentioned analysis, such as creep fracture time, it can be concluded that the oxidation layer rarely affects the creep fracture time of the ultraminiature specimen with AISI 304.

Due to its excellent high-temperature performance, AISI 304 has good oxidation resistance, and size effects are not observed in the material. Notwithstanding, a size effect or oxidation-induced reduction of the loading-bearing area in miniature creep tests has been reported. Thus, investigating the size effect using various materials, such as Ni-based superalloys and ferritic or martensitic heat-resistant steels, is the future goal of the research. Moreover, the improvement of the creep testing machine by adding an Ar gas chamber to prevent a potential decrease in the creep fracture time caused by oxidation is also a future objective.