**2. Materials and Methods**

#### *2.1. Materials*

In order to carry out the experimental work, a pipe with Ø48.3 mm and a thickness of 5 mm in SA 355 P91 steel provided by Vallourec Brazil (Belo Horizonte, Brazil) was selected. Its chemical composition and main mechanical properties provided by the supplier can be seen in Tables 1 and 2, respectively. The length of each sample was 150 mm, and the preparation (chamfer) was made in just one tip of the pipe. The preparation details can be seen in Table 3. Twenty-five samples have been produced under each set of conditions, in order to provide at least five samples for each kind of destructive test. The results of the tests carried out in this work present the average value followed by the standard deviation.


**Table 1.** Chemical composition of the parent metal SA 355 P91.

As filler metal, wire with Ø2 mm with the reference ER90S-B9 was used, which was provided by Electro Portugal, Lda. (Porto, Portugal); its chemical composition can be observed in Table 3. The mechanical properties of this filler metal follow the EN10204 standard. This filler metal was selected because it is recommended by the ASME manufacturing code for P91 steel.

#### *2.2. Methods*

Since the main objective of this work is to analyze the influence of different heat treatment cycles on P91 steel welded joints, the welding conditions were kept constant for producing the samples. Initially, the pipes were cut in a circular sawing machine Kaltenbach KKS 450 S, and the chamfer was performed in a conventional lathe Knuth V-Turn Pro (KNUTH Werkzeugmaschinen GmbH, Wasbek, Germany). After that, the samples were properly cleaned, avoiding the presence of chips. The welding process used in this work was GTAW in the position of 2G using a Kemppi (Kemppi, Lahti, Finland) Master TIG MLS 4000 multi-process power source welding machine. To perform the welding, the samples were properly aligned, and thermocouples were attached to the surface close to the weld joint. Following Kobelko's recommendation for this type of material (between 0.8 and 1.6 kJ/mm), a thermal delivery of 1.33 kJ/mm was selected, which conditioned the welding parameters. Good

engineering practice recommends that a current rating of about 45 A per each millimeter diameter of the filler material should be used. Thus, a current intensity of 95 A was selected, as the diameter of the wire used as the filler metal was ±2 mm. The main parameters used in the welding process are shown in Table 4.


**Table 4.** Main parameters used in the welding process.

The voltage was adjusted automatically, following the characteristic curve of the welding machine. After the welder obtained the current intensity, the voltage, and the range of thermal delivery, the traveling speed was adjusted, taking into account the welding position and the thickness of the pipe to be welded. For calculating the traveling speed, the following expression for the thermal delivery was used:

$$Q = k \times \left(\frac{lI \times I}{v}\right) \times 10^{-3} \tag{1}$$

where *Q* is the thermal delivery (kJ/mm), *k* is the thermal efficiency factor, *U* is the arc voltage (V), *I* is the current intensity (A), and *v* is the traveling speed (mm/s). This is valid for the first pass, because it is well known that for the second pass, a higher level of current can be used (+10%), as well as a lower traveling speed (−20%), in order to properly fulfill the chamfer. These new parameters for the second pass will result in higher thermal delivery, although it is controlled in order to prevent the appearance of welding defects.

For preheating, post-heating, and PWHT processes, a Weldotherm (Weldotherm, Essen, Germany) VAS 82-12 Digit 1000 Heat Treating Unit provided with 12-channel capacity was used, with six channels that were capable of using electrical resistances up to 135 A and another six channels that were capable of using electrical resistances up to 90 A.

Different thermal cycles have been applied to the welded samples in order to study the mechanical behavior of the joints. In Figure 1, it is possible to observe the different cycles selected, as well as the code used in this work, making the correlation with the subsequent analyses easier. Table 5 presents the main levels of temperature used in each set of heat treatments. The main argumentation behind the selection of the different thermal cycles imposed to the samples used in this study can be found below:


corrosion phenomena. In order to simulate this situation, preheating was used, but without post-heating treatment. At the end of the welding process, the samples have been protected with a ceramic fiber blanket, avoiding a fast cooling process until it reached room temperature. No phases' transformation time was used. The PWHT was applied just six months later;


**Figure 1.** *Cont.*

**Figure 1.** Different sets of thermal cycles used for each set of samples utilized in this work. Legend: (**a**) Pre-heating; (**b**) Welding; (**c**) Post-heating; (**d**) Transformation time; (**e**) Controlled cooling; (**f**) Post-welding heat-treatment (PWHT).


**Table 5.** Different thermal cycles applied to the P91 samples.

The purpose of this test is to verify to what extent hydrogen, if inserted into the weld, interferes with the joint together with the internal stress and corrosion in the weld. It is well known that the lack of tempering of this material after welding for extended periods can be detrimental. This is because, due to their late performance, the welds have a microstructure that is susceptible to internal stresses and often external stresses which, together with a certain level of hydrogen, can result in cold cracking, which is also known as hydrogen cracking. Apart from the presence of these agents, it is possible that welds may also be subject to corrosion, which may lead to stress cracking in addition to cold cracking. This was the main reason to include a waiting period of 6 months after the welding and before PWHT.

Figure 2 shows the setup used to perform the preheating process (a), as well as the welding process (b).

**Figure 2.** (**a**) Pipes preheating setup and (**b**) Pipe welding.

The characterization of the joints has been carried out following different techniques with a view to get important information about the properties achieved by the samples under each set of thermal cycles used. Thus, the characterization techniques used can be seen below, as well as the main reasons behind their selection.


After welding, the samples were immediately subjected to a first check in order to verify if the samples met the acceptance criteria defined in EN ISO 5817–Level C [33]. Afterwards, an additional visual inspection following ISO-17637:2003 standard [34] was performed, getting the corresponding approval. After that, samples were checked using liquid penetrant following the ISO 3452-1:2013 [35], looking for defects that had reached the surface. Since all the samples did not present surface signs of defects, they obtained the corresponding approval. Trying to detect internal defects, the samples also were subjected to X-ray inspection following the ISO 20769-2:2018 standard [36], using an ICM machine with 300 kVA, obtaining the corresponding approval as well. Indeed, some samples presented an excess of penetration, which is a defect that is acceptable, but there were no other defects such as root lack of penetration, porosities, or inclusions. The criteria used to approve the welds also followed ASME Code Sections I, V, VIII Div. 1 and 2, ASME B31.1 [37]. After these previous non-destructive tests, the samples were cut and machined in a vertical milling machine Baileigh VM-1054-3. A summary of these previous tests can be seen in Figure 3. After machining, it was necessary to smooth the weld until it is leveled with the base material, so that the entire test area was of the same thickness and section in order to remove the notch factor resulting from the border of the bead with the base material. This operation was performed for the tensile and bend samples.

In order to perform the other tests, different equipment was used, which is described as follows. Regarding the hardness assessment, three different tests were performed: hardness tests with portable equipment to carry out a first evaluation of the welded joints, a microhardness cross-section evaluation in order to assess the hardness reached in each zone of the joint, and finally another microhardness test allowing the detection of soft spots. Thus, the surface preparation was carried out using sandpaper provided with different grain sizes (80, 220, 310, 500, and 1000 mesh) until the Rz roughness parameter was lower than 5 μm. A Krautkramer–MIC 10 equipment provided with a MIC 205-A probe was used because it is portable and easy to handle, providing results accurate enough regarding the level required at this stage, which was just a general previous evaluation. The load used to perform these tests was 49 N (5 kgf) with a dwell time of 30 s, and a diamond Vickers indenter with 136◦ was used to perform the indentations. In order to obtain more accurate results, further microhardness tests were carried out using Shimadzu HSV-20 equipment, following the NP EN 1043-1 standard, using as the load 98.1 N (10 kgf) and a Vickers diamond indenter. However, to do that, it was necessary to improve the surface roughness through a new polishing process with diamond slurry of 1 μm over 10 min in order to decrease the Rz roughness to values under 2 μm. Two rows of indentations were produced as depicted in Figure 4. Finally, the opposite surface of the same samples was prepared following the same procedure as well, in order to identify soft spots close to the base material. Indeed, Newell [38]

argues that these soft points are responsible for a decrease in creep resistance by about 20% compared to an unaffected base material and are formed mainly in the fine grain region of the HAZ. Using the same microhardness equipment and indenter, these tests have been carried out using just 1 kgf (HV1) and a dwell time of 30 s, still following the ISO 6507:2018 standard [39]. Three rows of indentations were produced, one close to the weld root, another in the middle of the cross-section, and the last one close to the surface of the base material; all of the indentations were distanced 0.5 mm each other, which is the minimum distance recommended by the standard, avoiding the influence on the results among indentations. This value was selected based on a good compromise between the lowest value recommended for this effect and the small grain size usually observed close to the HAZ. In this case, just one sample was analyzed regarding each set of welding and heat-treatment conditions.

**Figure 3.** Different non-destructive test stages regarding the samples approval: (**a**) Liquid penetrant before welding; (**b**) Dimensional control; (**c**) Liquid penetrant after welding and cutting; (**d**) X-ray analysis.

**Figure 4.** Example on how the microhardness indentations were produced across the cross-section.

In order to analyze the chemical composition of the weld, an optical emission spectrometer Spectro, model Spectrolab M8 was used. This test was performed trying to understand the composition of the weld and corresponding dilution. The values achieved are based on six analyses in different samples, and the results presented are average values with standard deviations lower than 5%.

The micrographic analysis aims to analyze the microstructure present in the samples, thus allowing verification of the grain size and its distribution. In order to perform these analyses, a Carl Zeiss (Carl Zeiss, Oberkochen, Germany) optical microscope model Axioskop 2 Mat was utilized, using different lens and magnifications. This analysis was also an important tool to verify the influence of thermal cycles on the steel microstructure, both in melted zone (MZ) and the HAZ. This optical microscopy technique is ideal for microstructure analysis, grain size analysis, particle analysis, and the identification of voids or cracks. For this purpose, the samples were prepared by sanding and polishing with 80 mesh sandpapers, followed by 220, 310, 500, and finally 1000 mesh. Then, the samples were polished sequentially with 3 μm and 1 μm diamond slurry, after which the samples were etched using a reagent called Vilella, which consists of hydrochloric acid, 5 mL; picric acid, 2 g; and ethyl or methyl alcohol, 100 mL. The etching time was 20 s. The test was performed according to BS EN 1321:1997—Destructive test on welds in metallic materials—Macroscopic and microscopic examination of welds [40].

Tensile tests allow a vast characterization of the materials, both in terms of mechanical strength and ductility. Moreover, this type of test can be performed at different temperatures, simulating hard work conditions that can be applied to the materials and joints in service. In this case, tensile tests were performed at room temperature, 20 ◦C, and at high temperature, 600 ◦C, from which yield strength, rupture and elongation data were collected. In order to increase the confidence in the values collected, three tests were carried out for each condition considered in this work. In order to perform the tensile tests, as Instron universal testing machine model 4208 was used, which was provided with a load cell of 300 kN. Tensile test specimens were taken from the weld cross-section and prepared according to ASME IX: 2015–QW 150 [41]. Regarding the tensile tests carried out at elevated temperature (600 ◦C), they were performed in a similar universal testing machine (Instron 8562) provided with a heated camera where the samples are kept at constant temperature during the tensile tests. The load cell used is a 100 kN cell. The samples were prepared following the ISO 6892-2:2018 standard [42].

Bending tests are usual in welded samples because they allow the assessment of the samples' behavior under very demanding work circumstances. In this work, bending tests were performed using a hydraulic CIATA press, model P-115/HP provided with a maximum compression capacity of 147 kN (15 tons). Samples were tested according to ASME IX–QW163 [43], with a bending angle of 180◦, a bearing distance of 36 mm, and a spindle diameter of 24 mm. Samples were used in which the compressive effort was performed on the welding face and others were performed on the welding root to cover all situations. After the tests, the samples were analyzed using the penetrant liquid technique, which was previously described in this work.

#### **3. Results and Discussion**
