The Effect of Material Heterogeneity and Temperature on Impact Toughness and Fracture Resistance of SA-387 Gr. 91 Welded Joints

This paper presents the analysis of the behavior of welded joints made of 9–12% Cr-Mo steel SA-387 Gr. 91. The successful application of this steel depends not only on the base metal’s (BM) properties but even more on heat-affected-zone (HAZ) and weld metal (WM), both at room and at operating temperature. Impact testing of specimens with a notch in BM, HAZ, and WM was performed on a Charpy instrumented pendulum to enable the separation of the total energy in crack-initiation and crack-propagation energy. Fracture toughness was also determined for all three zones, applying standard procedure at both temperatures. Results are analyzed to obtain a deep insight into steel SA 387 Gr. 91’s crack resistance properties at room and operating temperatures. Results are also compared with results obtained previously for A-387 Gr. B to assess the effect of an increased content of Chromium.


Introduction
The Cr-Mo steel SA-387 Gr. 91 belongs to a group of heat-and creep-resistant 9-12% Cr steels. They are introduced into practice to replace 2.25% Cr-Mo steel for operating temperatures above 565 • C, with the maximum service temperature of Gr. 91 equaling circa 600 • C [1].
The Cr-Mo steel SA-387 Gr. 91 has exceptional mechanical properties, including crack initiation and propagation resistance, making it an excellent choice for pressure vessels operating at elevated temperatures. It is a simple matter to compare SA-387 Gr. 91, in its role as a base metal, with more conventional steels, such as SA-387 Gr. B, and to find out that conventional design methods will lead to significant reductions in pressure vessel thickness and costs in general [1]. However, having in mind the importance and complexity of welded joints, it is of utmost importance to obtain a deep insight into the behavior of all zones (base metal, weld metal, heat-affected zone) to ensure the safe application and exploitation of Cr-Mo steel. Knowing that welded joint crack sensitivity increases, the more complex the composition and structure of a steel becomes, it is reasonable to assume that a complete overview of SA 387 Gr. 91 application must include a detailed analysis of its crack resistance in all welded joint zones. This should include at least Charpy toughness and fracture toughness testing, both at room and elevated temperatures, up to 575 • C, which is the aim of this research.
In a limited number of papers published about the effect of material heterogeneity and temperature on steel SA-387 Gr. 91's behavior, most of the focus was on strength and creep on material heterogeneity and temperature effects, as already briefly outlined in [19], but also with a focus on the Chromium effect, which was not previously analyzed in this way. Therefore, the results presented here will not be only analyzed on their own, but also in comparison with corresponding results for A-387 Gr. B to obtain a better insight into the effect of a significantly increased content level of Chromium. In any case, the focus in this research is on the effect of weldment heterogeneity (i.e., the different properties of BM, WM, and HAZ) and Cr content on the crack resistance of a welded joint.

Methods
Steel SA-387 Gr. 91 is designed to have the minimum yield stress of 450 MPa and minimum impact energy of 41 J at room temperature, with the idea that is will be able to work at elevated temperatures with sufficient strength and toughness [1,19]. For this research, steel SA-387 Gr. 91, thickness 15 mm, produced in "Steelwork ACRONI" Jesenice Slovenia, was used, with the chemical composition shown in Table 1.
Welding was performed in 4 root and 10 filler passes, using a Gas Tungsten Arc Welding (GTAW) with BOEHLER C9 MV-IG Ø2.4 mm filler metal to ensure high quality of root passes 1-4 and a Shielded Metal Arc Welding (SMAW) with BOEHLER FOX C9 MV electrode, diameters 2.5 (passes 5-9) and 3.25 mm (passes [10][11][12][13][14], as shown schematically in Figure 1. The chemical compositions and mechanical properties of filler metals are shown in Table 2. Welding parameters and linear energies (as shown in Table 3) were chosen carefully to adjust cooling speed and optimize welded-joint microstructure, with thermal efficiency coefficients taken as 0.6 (GTAW) and 0.8 (SMAW).
Pre-heating was performed at 250 • C, while the inter-pass temperature was 200-300 • C. Post-Weld Heat Treatment (PWHT) was applied, consisting of tempering at 250 • C, followed by heating up to 750 • C (rate 100-150 • C/h), holding at 750 for 2 h, and cooling down to 400 • C (rate 150 • C/h), with final cooling at the still air.

Impact Testing on Charpy Instrumented Pendulum
The testing procedure was applied according to SRPS EN ISO 9016:2013 [20], including specimen shape and size, as well as notch V-2 position, Figure 2. Testing was performed on an instrumented Charpy pendulum SCHENCK TREBELL 150/300 J at room temperature, 20 °C, and an elevated temperature, 575 °C. The higher temperature was chosen as it is a common service temperature for this steel, whereas room temperature was used as a reference, so that the effect of high temperature on the steel could be evaluated. Three specimens were extracted from each characteristic zone with the crack tip positioned in the BM, WM, and HAZ. In the case of the BM, the specimens were taken from a location far from the weld metal, as a common practice to avoid welding heat effects. In the case of the HAZ, the specimen tip was located in the HAZ, as close to the WM as possible (as shown in Figure 2), since the CGHAZ was shown to have the lowest toughness in HAZ [6]. In the case of the WM specimen, the tip was located close to the center line. Since the tests were performed using an instrumented Charpy pendulum, it was possible to separate crack initiation and propagation energies, and to evaluate the effect of the notch location on the impact properties and plasticity. In this way, it was possible to determine the energy required for initiating a crack and the energy required for its propagation, enabling better understanding of the crack resistance of tested material, as explained in more details in [21], including different methods to separate these two energies. In this research, separation was performed according to the force maximum value, so that the area to the left represents the energy for crack initiation, Ai, the area to the right the energy for crack propagation, Ap, Figure 3.

Impact Testing on Charpy Instrumented Pendulum
The testing procedure was applied according to SRPS EN ISO 9016:2013 [20], including specimen shape and size, as well as notch V-2 position, Figure 2. Testing was performed on an instrumented Charpy pendulum SCHENCK TREBELL 150/300 J at room temperature, 20 • C, and an elevated temperature, 575 • C. The higher temperature was chosen as it is a common service temperature for this steel, whereas room temperature was used as a reference, so that the effect of high temperature on the steel could be evaluated. Three specimens were extracted from each characteristic zone with the crack tip positioned in the BM, WM, and HAZ. In the case of the BM, the specimens were taken from a location far from the weld metal, as a common practice to avoid welding heat effects. In the case of the HAZ, the specimen tip was located in the HAZ, as close to the WM as possible (as shown in Figure 2), since the CGHAZ was shown to have the lowest toughness in HAZ [6]. In the case of the WM specimen, the tip was located close to the center line.

Impact Testing on Charpy Instrumented Pendulum
The testing procedure was applied according to SRPS EN ISO 9016:2013 [20], including specimen shape and size, as well as notch V-2 position, Figure 2. Testing was performed on an instrumented Charpy pendulum SCHENCK TREBELL 150/300 J at room temperature, 20 °C, and an elevated temperature, 575 °C. The higher temperature was chosen as it is a common service temperature for this steel, whereas room temperature was used as a reference, so that the effect of high temperature on the steel could be evaluated. Three specimens were extracted from each characteristic zone with the crack tip positioned in the BM, WM, and HAZ. In the case of the BM, the specimens were taken from a location far from the weld metal, as a common practice to avoid welding heat effects. In the case of the HAZ, the specimen tip was located in the HAZ, as close to the WM as possible (as shown in Figure 2), since the CGHAZ was shown to have the lowest toughness in HAZ [6]. In the case of the WM specimen, the tip was located close to the center line. Since the tests were performed using an instrumented Charpy pendulum, it was possible to separate crack initiation and propagation energies, and to evaluate the effect of the notch location on the impact properties and plasticity. In this way, it was possible to determine the energy required for initiating a crack and the energy required for its propagation, enabling better understanding of the crack resistance of tested material, as explained in more details in [21], including different methods to separate these two energies. In this research, separation was performed according to the force maximum value, so that the area to the left represents the energy for crack initiation, Ai, the area to the right the energy for crack propagation, Ap, Figure 3. Since the tests were performed using an instrumented Charpy pendulum, it was possible to separate crack initiation and propagation energies, and to evaluate the effect of the notch location on the impact properties and plasticity. In this way, it was possible to determine the energy required for initiating a crack and the energy required for its propagation, enabling better understanding of the crack resistance of tested material, as explained in more details in [21], including different methods to separate these two energies. In this research, separation was performed according to the force maximum value, so that the area to the left represents the energy for crack initiation, A i , the area to the right the energy for crack propagation, A p , Figure 3.

Fracture Toughness, KIc, Testing
Three-point single-edge bending (SEB) specimens were used for fracture toughness, KIc, measurement at room temperature, whereas modified CT specimens were used at elevated temperature, 575 °C (as shown in Figure 4). This modification was needed due to the shape of the chamber used for testing at 575 °C and had no effect on fracture toughness values, since the stress-strain state at the crack tip was not affected. Fracture toughness, KIc, was determined via critical J integral, JIc, applying rules of elastic-plastic fracture mechanics (EPFM) [22]: where E is the Elasticity modulus, and is ν the Poisson ratio. Standard procedure is defined in the ASTM 1820 standard [23] along with the specific aspects for a welded joint testing set out in [24]. A crack was produced on the HF testing machine, and its length was measured after the experiment, as defined in [23].

Fracture Toughness, K Ic , Testing
Three-point single-edge bending (SEB) specimens were used for fracture toughness, K Ic , measurement at room temperature, whereas modified CT specimens were used at elevated temperature, 575 • C (as shown in Figure 4). This modification was needed due to the shape of the chamber used for testing at 575 • C and had no effect on fracture toughness values, since the stress-strain state at the crack tip was not affected. Fracture toughness, K Ic , was determined via critical J integral, J Ic , applying rules of elastic-plastic fracture mechanics (EPFM) [22]: where E is the Elasticity modulus, and is ν the Poisson ratio. Standard procedure is defined in the ASTM 1820 standard [23] along with the specific aspects for a welded joint testing set out in [24]. A crack was produced on the HF testing machine, and its length was measured after the experiment, as defined in [23].

Fracture Toughness, KIc, Testing
Three-point single-edge bending (SEB) specimens were used for fracture toughness, KIc, measurement at room temperature, whereas modified CT specimens were used at elevated temperature, 575 °C (as shown in Figure 4). This modification was needed due to the shape of the chamber used for testing at 575 °C and had no effect on fracture toughness values, since the stress-strain state at the crack tip was not affected. Fracture toughness, KIc, was determined via critical J integral, JIc, applying rules of elastic-plastic fracture mechanics (EPFM) [22]: where E is the Elasticity modulus, and is ν the Poisson ratio. Standard procedure is defined in the ASTM 1820 standard [23] along with the specific aspects for a welded joint testing set out in [24]. A crack was produced on the HF testing machine, and its length was measured after the experiment, as defined in [23].

Impact Testing
Results of the impact tests are provided for BM, WM, and HAZ in Tables 4-6 ness at room temperature is the highest in BM, closely followed by HAZ. High resistance to cracking in HAZ is even more pronounced when energy components are considered, since it has the highest resistance to crack initiation. In any case, one should keep in mind that all zones in SA 387 Gr. 91 have a relatively high impact energy, both for crack initiation and propagation, making their welded joints resistant to cracking. At this point, one should notice that such result actually leads to the conclusion that the welding procedure specification for SA 387 Gr. 91 is well defined, and welding itself is well performed.
The results of the impact testing are in good agreement with the presented microstructures and hardness values, since the highest impact energy (BM) corresponds with the lowest hardness, and the lowest impact energy (WM) corresponds with the highest hardness.

Impact Testing
Results of the impact tests are provided for BM, WM, and HAZ in Tables 4-6, respectively. Characteristic examples of F-t diagrams are shown in Figures 5-7 for BM, WM, and HAZ, respectively. As one can see from the results presented in Tables 4-6, impact toughness at room temperature is the highest in BM, closely followed by HAZ. High resistance to cracking in HAZ is even more pronounced when energy components are considered, since it has the highest resistance to crack initiation. In any case, one should keep in mind that all zones in SA 387 Gr. 91 have a relatively high impact energy, both for crack initiation and propagation, making their welded joints resistant to cracking. At this point, one should notice that such result actually leads to the conclusion that the welding procedure specification for SA 387 Gr. 91 is well defined, and welding itself is well performed.
The results of the impact testing are in good agreement with the presented microstructures and hardness values, since the highest impact energy (BM) corresponds with the lowest hardness, and the lowest impact energy (WM) corresponds with the highest hardness.           Testing at the operating temperature indicates similar behavior, since the reduction of energies is similar: BM 29-43%, WM 26-47%, and HAZ 9-43%. Therefore, energy values at the operating temperature, compared with the room temperature are as follows: BM 57-71%, WM 53-74%, HAZ 57-91%, which are still relatively high. The lowest value is energy for crack initiation, A I = 28 J, which was recorded in WM and is still reasonable from a practical point of view.
The effect of different zones in a welded joint on crack initiation and propagation is also visible in Figures 8-10, where fractographies of BM, WM, and HAZ are shown, respectively, for both testing temperatures. It is clear that only Figures 8a, 9a and 10a, which present the crack initiation process at 575 • C, do not show completely ductile fracture surfaces, which is in agreement with the lower energies recorded for crack initiation in these specimens (42, 28, and 40 J, respectively). However, they do not represent brittle fractures either; all fractographies indicate sufficient toughness values and high resistance to crack initiation and propagation. Moreover, one should notice relatively small differences in crack initiation and propagation energies between the different zones in a welded joint made of SA387 Gr. 91, which is also proved by the presented fractographies.  Testing at the operating temperature indicates similar behavior, since the reduction of energies is similar: BM 29-43%, WM 26-47%, and HAZ 9-43%. Therefore, energy values at the operating temperature, compared with the room temperature are as follows: BM 57-71%, WM 53-74%, HAZ 57-91%, which are still relatively high. The lowest value is energy for crack initiation, AI = 28 J, which was recorded in WM and is still reasonable from a practical point of view.
The effect of different zones in a welded joint on crack initiation and propagation is also visible in Figures 8-10, where fractographies of BM, WM, and HAZ are shown, respectively, for both testing temperatures. It is clear that only Figures 8a, 9a and 10a, which present the crack initiation process at 575 °C, do not show completely ductile fracture surfaces, which is in agreement with the lower energies recorded for crack initiation in these specimens (42, 28, and 40 J, respectively). However, they do not represent brittle fractures either; all fractographies indicate sufficient toughness values and high resistance to crack initiation and propagation. Moreover, one should notice relatively small differences in crack initiation and propagation energies between the different zones in a welded joint made of SA387 Gr. 91, which is also proved by the presented fractographies.    The results of the impact testing for A387 Gr. B, obtained at room temperature, are presented in Tables 7-9 for BM, WM, and HAZ, respectively. The distribution of energies, both total and separated, is similar as for 9% Cr steel, the highest values are in BM, but are followed closely by both HAZ and WM, in this case. These results also lead to the conclusion that the welding procedure specification for A 387 Gr. B is well defined, and the welding itself is well performed.
The reduction of energy at operating temperature is smaller for steel with 1% Cr than for steel with 9% Cr, with similar distribution of energy: BM 18-38%, WM 8-30%, HAZ 20-29%. The levels of energy in relation to the room temperature are: BM 62-82%, WM 70-92%, HAZ 71-80%. The lowest individual value for initial energy, A I = 38 J, is recorded in BM and is still satisfactory. Nevertheless, one should not forget that the operating temperature for 1% Cr steel is 540 • C, i.e., lower than that for 9% Cr (575 • C), so the reduction of energies was expected, not only because of the simpler microstructure (less Cr).

Fracture Toughness Testing
Fracture toughness values are obtained via J Ic , as explained in Section 2.2, using J-R curves as shown in Figures 11-13 for characteristic examples of BM, WM, and HAZ testing, respectively. Calculated K Ic values for SA 387 Gr. 91 steel are provided in Tables 10-12 for BM, WM, and HAZ, respectively, clearly indicating that the K Ic values are satisfactory, with the highest values in BM (175.0 and 124.4 J for the room and operating temperature, respectively), the lowest in WM (125.7 and 91.1 J), and in-between in HAZ (146.4 and 111.9 J). The effect of heterogeneity and temperature is similar, as in the case of impact toughness, with HAZ being somewhat more sensitive to cracking, and with a slightly smaller reduction of K Ic values (the ratio between 20 and 575 • C values is circa 1.3 compared to circa 1.6 for impact toughness) with increased temperature.

Fracture Toughness Testing
Fracture toughness values are obtained via JIc, as explained in Section 2.2, using J-R curves as shown in Figures 11-13 for characteristic examples of BM, WM, and HAZ testing, respectively. Calculated KIc values for SA 387 Gr. 91 steel are provided in Tables 10-12 for BM, WM, and HAZ, respectively, clearly indicating that the KIc values are satisfactory, with the highest values in BM (175.0 and 124.4 J for the room and operating temperature, respectively), the lowest in WM (125.7 and 91.1 J), and in-between in HAZ (146.4 and 111.9 J). The effect of heterogeneity and temperature is similar, as in the case of impact toughness, with HAZ being somewhat more sensitive to cracking, and with a slightly smaller reduction of KIc values (the ratio between 20 and 575 °C values is circa 1.3 compared to circa 1.6 for impact toughness) with increased temperature.         As in the case of impact toughness, one should notice relatively small differences in K Ic values between the different zones in welded joints made of SA387 Gr. 91, proved also by the presented fractographies (as shown in Figures 14-16), indicating sufficiently ductile material. Even in the case of WM at 575 • C (as shown in Figure 15b), which appears to be a brittle fracture, it was found that this is actually a 'local brittle zone' (LBZ), not uncommon for WM, especially if made of alloyed steel. The same fractography, but with a magnification of 200×, is shown in Figure 17, also indicating some typical features of a ductile fracture. As in the case of impact toughness, one should notice relatively small differences in KIc values between the different zones in welded joints made of SA387 Gr. 91, proved also by the presented fractographies (as shown in Figures 14-16), indicating sufficiently ductile material. Even in the case of WM at 575 °C (as shown in Figure 15b), which appears to be a brittle fracture, it was found that this is actually a 'local brittle zone' (LBZ), not uncommon for WM, especially if made of alloyed steel. The same fractography, but with a magnification of 200 ×, is shown in Figure 17, also indicating some typical features of a ductile fracture.     Tables  13-15 for BM, WM, and HAZ, respectively. In this case, the results for fracture toughness are different from those for impact toughness in two important aspects: the first is that 1% Cr steel has lower values than 9% Cr, and the second is that WM is now the region with the highest crack resistance, whereas HAZ is the weakest link. However, the differences are very small, since the average values for WM are circa 10% higher, and for HAZ circa 10% lower, than BM average values. Therefore, the effect of heterogeneity is less pronounced than in the case of SA 387 Gr. 91 steel, whereas the effect of temperature is slightly stronger (the ratio between 20 °C and 575 °C values is circa 1.4, which is almost the same as in the case of SA 387 Gr. 91).    Tables  13-15 for BM, WM, and HAZ, respectively. In this case, the results for fracture toughness are different from those for impact toughness in two important aspects: the first is that 1% Cr steel has lower values than 9% Cr, and the second is that WM is now the region with the highest crack resistance, whereas HAZ is the weakest link. However, the differences are very small, since the average values for WM are circa 10% higher, and for HAZ circa 10% lower, than BM average values. Therefore, the effect of heterogeneity is less pronounced than in the case of SA 387 Gr. 91 steel, whereas the effect of temperature is slightly stronger (the ratio between 20 °C and 575 °C values is circa 1.4, which is almost the same as in the case of SA 387 Gr. 91).  Tables 13-15 for BM, WM, and HAZ, respectively. In this case, the results for fracture toughness are different from those for impact toughness in two important aspects: the first is that 1% Cr steel has lower values than 9% Cr, and the second is that WM is now the region with the highest crack resistance, whereas HAZ is the weakest link. However, the differences are very small, since the average values for WM are circa 10% higher, and for HAZ circa 10% lower, than BM average values. Therefore, the effect of heterogeneity is less pronounced than in the case of SA 387 Gr. 91 steel, whereas the effect of temperature is slightly stronger (the ratio between 20 • C and 575 • C values is circa 1.4, which is almost the same as in the case of SA 387 Gr. 91).

Conclusions
Based on the results presented in this paper, one can conclude the following: • Both steels, SA-387 Gr. 91 and A-387 Gr. B, as well as their welded joints, have high resistance to cracking, both for static and impact loading. This conclusion also holds for SA 387 Gr. 91 WM, even though its resistance to cracking is lower than BM and HAZ, but well above 41 J, which is the minimum value for the BM.

•
The effect of material heterogeneity on impact toughness is more heavily expressed for SA-387 Gr. 91 than for A-387 Gr. B, since the WM in the former case has lower values of crack initiation and growth energies, whereas these values are balanced in the latter case. A reduction of impact toughness in the case of SA-387 Gr. 91 steel is mostly due to crack-growth energy, which is significantly smaller than for SA-387 Gr. 91 BM and HAZ, but still at a satisfying level.

•
The effect of temperature on impact toughness is similar, but more pronounced, since both energies are lower in all cases, approximately 1/3 less than at room temperature, but still at a satisfying level.

•
The effect of material heterogeneity on fracture toughness is similar to its effect on impact toughness, but more expressed for SA-387 Gr. 91 than for A-387 Gr. B, for the same reason as in the case of impact toughness. The effect of temperature on fracture toughness is also similar to its effect on impact toughness. One can say that the behavior of both materials and their welded joints in respect to cracking is practically the same for static and impact loading.