*Article* **Evaluation of Welded Joints in P91 Steel under Di**ff**erent Heat-Treatment Conditions**

**Francisco José Gomes Silva 1,2,\*, António Pedro Pinho 1, António Bastos Pereira <sup>2</sup> and Olga Coutinho Paiva <sup>1</sup>**


Received: 18 December 2019; Accepted: 7 January 2020; Published: 8 January 2020

**Abstract:** P91 steel has been of interest to many researchers over the past two decades. This interest is because this steel has very interesting characteristics for application in power plants, where it is common to have pipes that need to support steam at temperatures between 570 and 600 ◦C, and at pressures in the range of 170 to 230 bar. These working conditions are quite severe for most common steels, requiring increased high-temperature mechanical strength as well as high creep resistance. The manufacture of these pipes normally includes welding operations, which must preserve the main characteristics of this type of steel. This justifies the concern of the researchers to ensure the best welding conditions so that the preservation of the properties of these steels becomes possible. The present work intends to depict the best results obtained varying the heat-treatment conditions applied to weldments made on heat-resistant steel P91. This steel usually takes the designation SA 213 T91 (seamless tube) or SA 335 P91 (seamless pipe), according to ASME II, as well as the designation X10CrMOVNb9-1 according to EN 10216-2. The purpose of this study is to compare the behavior of pipe welding under different post-welding heat-treatment (PWHT) conditions. One of them is performed with thermal cycles (preheating, post-heating, and the post-weld heat treatment) in agreement with most construction codes and standard rules. The second one is performed without any thermal cycle before and after welding. Both welds were made by the same process, TIG (Tungsten Inert Gas, or GTAW—Gas Tungsten Arc Welding) in the horizontal position (2G according to ASME IX) and the same welding parameters. In order to evaluate the results obtained in the welds, microstructure analyses, hardness measurements, bending tests, and tensile tests at room and high temperature (600 ◦C) have been performed. Other tests were also carried out according to the quality procedures, such as visual, penetrant dye, and X-ray tests. Regarding the different strategies used in the heat treatments, the best results have been obtained using a strategy similar to the one currently in use and recommended by construction codes and steel manufacturers but excluding the phases' transformation time, and it was possible to observe that the tensile strength is impaired by about 2% to 9% at room and elevated temperatures, respectively; the elongation is reduced by 39% at room temperature but keeps a good performance at elevated temperature; the hardness profile is very similar at both temperatures; the microstructure presented is compatible with the requirements; and no cracking trend has been reported. Thus, a new strategy for the welding heat treatment of grade 91 steels was drawn, saving energy and processing time.

**Keywords:** P91 steel; heat-resistant steels; welding; heat treatment; PWHT; welds characterization; microstructures; tensile strength; hardness; heat-treatment processing time; sustainability

#### **1. Introduction**

Grade 91 steel was developed based on steels that emerged in the 1960s with typically 12% Cr content. The development of grade 91 was mainly due to creep problems, as 12% Cr steels traditionally failed when exposed to prolonged creep conditions such as those in power plants where these steels were used, exposing them to high pressures and temperatures of around 565 ◦C. Thus, steels with increased creep resistance have emerged. Grade 91 was originally developed by Oak Ridge National Laboratory in Tennessee, USA, typically consisting of 9% Cr and 1% Mo, which were initially called P9 steel presented as its main focus use in power plants [1]. Subsequently, this steel was studied and its composition was evolved through the addition of other elements, such as vanadium and Nb, and with controlled N content, thus giving rise to 91 steel grade. This new grade of steel substituted the P22 steel grade and can assume various designations, unfolding under designations such as SA 213 P91 (seamless tube) and SA 335 P1 (seamless pipe) according to ASTM, or X10CrMoVNb9-1 according to BS EN 10216-2 [2]. The P91 steel grade also responded to the demand for the increased efficiency of power plants, which now need to operate at higher temperatures in order to release lower amounts of CO2 for the same volume of energy generated. Indeed, the latest composition of the P91 steel grade allows continuous working at temperatures in the range of 600 ◦C without being affected by creep phenomena, even under elevated stress conditions. Since welding of the various components that form part of the power plants is required, P91 steel has been the subject of numerous studies, most of which are briefly described below. The excellent properties of P91 steel also come from the careful distribution of fine Nb and V carbonitrides, which have a microstructure that can be changed during the welding process [3].

Several researchers have devoted their attention to the characterization of P91 steel welds made using different processes. In a recent study, Vidyarthy and Dwivedi [4] compared the use of TIG and A-TIG (Activated Flux TIG) processes (GTAW) on P91 steel welds, investigating the influence of CeO2 and MoO3-based activating flux on some factors that strongly condition welding, such as such as heat input, weld bead geometry, and angular distortion during single-pass execution. The use of A-TIG aims to overcome the productivity limitations imposed by the conventional TIG process, which is essentially geared to small thicknesses. The thin activated flow layer used in the A-TIG process substantially improves the process productivity [5]. Welding beads performed by both processes were also investigated, analyzing the resulting microstructure, mechanical strength, microhardness, and impact strength (Charpy). It was observed that for the same set of parameters used in both welding processes, the A-TIG process promoted an increase in the heat input transmitted to the joint due to the action of the activating fluxes, which resulted in an increase of 200% in the joint penetration when CeO2 flux is used, and 300% when MoO3 flux is used, compared to the conventional TIG process. In the analysis of welded joints, other benefits were also observed, such as less angular distortion, which dropped from 1.96◦ in conventional TIG welding to 0.78◦ using CeO2 flux or 0.12◦ when using MoO3 flux. The microstructure was predominantly dominated by martensite in the welding zone, and coarse precipitates such as M23C6 carbides were also observed in the primary austenite grain boundaries. The ultimate tensile strength (UTS) in welded samples increased by 2% related to the parent metal, and the microhardness increased as well. Conversely, impact resistance decreased in the A-TIG process compared to the conventional TIG process. Dhandha and Bandheka [5] also studied the A-TIG process applied to P91 steel using as fluxes Fe2O3, ZnO, MnO2, and CrO3. These authors also confirmed that penetration is always improved at least 100% using these kind of activated fluxes in that process, and a decrease in bead width was observed as well, as usually required. The best results were achieved using ZnO activated flux. The surface appearance/morphology of the weldments was considered as very good.

Marzocca et al. [3] studied P91 steel welds performed by means of the flux cored arc welding (FCAW) process, using two different rutilic filler metals (E91T1 and E91T1-G) and 80% Ar/20% CO2 shielding gas. The main focus of that work was to study the resulting microstructure of the welded zone, using five welding passes to fill up the chamfer previously prepared and a heat input energy of 1.5 kJ/mm. M23C6 carbides were found in all zones, i.e., the parent metal (PM), fused zone (FZ), and heat-affected zone (HAZ). VN (Vanadium-based) precipitates were also observed in all zones but with a decreased size and greater dispersion. However, NbCN was only found in the PM and HAZ. In addition, trying to overcome the lack of productivity characteristic of the conventional TIG process, Krishnan et al. [6] used the pulsed gas arc welding (GMAW-P) process using a cored wire filler material to weld 12 mm thick P91 steel sheets in a single pass. The authors reported that the best welding results were obtained using 75◦ bevel aperture and 1.38 kJ/mm heat input, which corresponds to a welding speed of 320 mm/min and a current intensity of 270 A. Very interesting properties of mechanical strength (UTS = 812–849 MPa) and the impact strength of the weld bead (104–127 J) have also been reported. The deposition rate with flux cored wire was increased by about 42% compared to the use of a common solid wire, considering the same set of parameters and welding conditions. Other authors using the same process reported a significant decrease in the defects generated during the welding process, namely spatter, welding porosity, and lack of fusion decreasing, as well the weld width [7,8]. Other processes such as shielded metal arc welding (SMAW), submerged arc welding (SAW), and flux cored arc welding (FCAW) have also been tested to maximize the welding efficiency of P91 steels, but non-metallic inclusions have been observed in the weld beads, loss of toughness, and excessively high oxygen content in the welds, considering the studies conducted during the 2000–2009 decade [9,10]. However, a further study developed by Arivazhagan and Kamaraj [11] in 2015 is in line with other works published more recently, which reported a very low amount of fine microinclusions less than 2 μm in size, allowing toughness values around 47 J. However, this value can be improved by 15%–25% using 100% Argon instead 80% Ar/20% CO2. An increase in the post-welding heat-treatment (PWHT) duration at 760 ◦C from 2 to 5 h has been reported as the main factor behind the 30% to 50% toughness improvement in welded joints.

Shanmugarajan et al. [12] used an autogenous laser beam in P91 steel welding but reported the presence of the delta ferrite phase in the weld beads, which was attributed to the amount of heat given to the joint. Kundu et al. [13] used an electron beam to weld P91 steel, but too high residual stresses were observed in welding of thicker thicknesses. In order to increase the efficiency in the GTAW process, Pai et al. [14] reached an increased filler metal deposition rate using it in preheated conditions, but in contrast, they observed a significant decrease in the joint toughness, which was attributed to the excessive heat passed to the joint in the process.

One of the main problems that has been worrying researchers is the possible drop in toughness and creep behavior due to the welding process. This has motivated several studies. El-Dosoky et al. [15] investigated the behavior of welded joints subjected to prolonged exposure at 600 ◦C under 120 MPa and 70 MPa loads, verifying that the creep resistance is conditioned by the fine grain of the heat-affected zone (FGHAZ). It has also been reported that welded samples have a higher creep rate than the parent metal mainly in the tertiary region, and that creep begins earlier in welded samples relative to the parent metal for the same load applied at 600 ◦C. Hyde et al. [16] studied the crack growth in welded P91 steel samples by creep crack growth tests at 650 ◦C, using compact strain (CT) test specimens, comparing the results obtained experimentally with simulations performed by the finite element method (FEM). The authors also confirmed a good correlation between the creep crack growth rates in the P91 parent metal and the cross-weld specimens for a given contour-integral (C\*) [17], in which the crack growth rate was 10 times higher in the cross-welded CT specimens than those of the parent metal. Using the same type of specimens, Kumar et al. [18] reported similar conclusions, adding the idea that the HAZ enables the faster generation and growth of cracks to the detriment of the parent material and melted zone, creating the best conditions for the deviation of crack paths from those zones to the HAZ. In addition, using CT specimens of P91 steel welded and non-welded, Venugopol et al. [19] reached similar conclusions, using the fracture mechanics parameter C\*. The conclusion drawn by these researchers, which was later confirmed by other works already cited here, emphasized that the creep crack growth rate is higher at the HAZ, which is especially true when lower C\* values are reported. A similar study was recently conducted by Baral et al. [20], using non-welded P91 steel samples and welded samples of the

same material, in a temperature range between 600 and 650 ◦C, and loads between 50 and 180 MPa. The observations made by these authors allowed ratifying the previous opinion of other researchers, in which the influence of the Cr23C6 coarse grains in the HAZ intercritical zone is clear, significantly conditioning the creep resistance, and that the minimum creep rate clearly follows Norton's power law. Trying to minimize the HAZ width, Divya et al. [21] carried out a comparative study between the laser welding (LW) of P91 steel welding, which leaves a HAZ of about 1 mm, with shielded metal arc welding (SMAW), which leaves a HAZ of about of 2.5 mm. The study was conducted to study the failure under creep conditions in the HAZ, which is commonly known as Type IV cracking. It was found that the microstructural damage induced by the LW is lower than in the case of the SMAW process. Although the width of the HAZ was effectively smaller in the case of LW, all specimens failed in the intercritical region of the HAZ. This shows that there is no benefit brought by LW in this case, because no significant improvements are brought to the Type IV cracking resistance. Wang et al. [22] performed PWHT at different temperatures on P91 welds, noting that there is a transition from Type IV cracking to Type I cracking in the HAZ intercritical region when moving from 600 to 840 ◦C treatment temperatures. In fact, the authors emphasize the idea that it is impossible to eliminate the HAZ intercritical region by PWHT if the temperature at which this treatment is performed does not exceed the Ac1 critical temperature of the parent metal. The intercritical structure formed on the basis of transformed austenite grains and untransformed ferrite grains in the fused zone may be the source of creep failure, i.e., Type IV cracking. These conclusions emphasize the importance of the temperature at which the PWHT must be conducted. Thus, PWHT conditions deserve special attention.

Sharma et al. [23] recently studied the effect of PWHT on welded P91 steel, reporting that the best conditions encountered to maximize the mechanical strength of welded pipe would be a 2 h treatment at 760 ◦C. In this study, the TIG process with heated wire was used, which allowed a smaller heat passage to the welded joint, minimizing the HAZ. In 2014, Venkata et al. [24] reported that the maximum temperature at which a PWHT should be performed is 770 ◦C and should always be lower than the austenite starting temperature (Ac1). In a study conducted in 2012 by Paddea et al. [25], the highest residual stresses (600 MPa) have been reported to be located near the outer boundary of the HAZ and toward the weld root in both PWHT and as-welded samples. As a result of these residual stresses, premature Type IV creep failures were observed in these 9–12% Cr (P92) steel welds. However, after PWHT, the residual stresses dropped to values around 50 MPa in the vicinity of the HAZ. Regardless of PWHT, the region where the highest level of residual stresses was measured has always been HAZ's intercritical region, which is the most vulnerable to Type IV cracking phenomena. Pandey et al. [26] performed creep tests at a temperature of 620 ◦C and loaded in the range of 150 to 200 MPa in multi-pass welded samples in P91 steel both in an as-welded condition and subjected to a set of heat treatments after welding. This set of treatments consisted of keeping the samples at 760 ◦C for 2 h with subsequent air cooling, followed by a re-austenitization treatment at 1040 ◦C for 1 h, and then further tempering at 760 ◦C again for 2 h, with a new air cooling process. It was reported by the authors that this treatment substantially increased the creep life of the samples, especially for a 150 MPa load. Laves phases have also been reported, as well as the change in the preferred failure location which, when in the as-welded condition, was caused by Type IV cracking in the HAZ intercritical region, but when subjected to the latter treatment, the breaking zone happened in the base material because the treatment conveniently unifies the structure along the samples. The same results are also reported by some of the same authors in [27]. Similar experiments carried out by a similar team of authors but using multi-pass shielded arc welded metal (SMAW) in P91 steel butt joints on 18 mm thick plate samples, which showed that the hydrogen taken to the joint in the welding process (6.21 mL/100 g) gave rise to hydrogen embrittlement, which is a situation that cannot be overcome by the sequence of treatments performed (identical to that described above). However, without such a large content of hydrogen present, the treatment produces a clear uniformity of the microstructure in the welded samples [28].

Several studies focusing on welding P91 steel with other materials, i.e., dissimilar joint, including P92 steel [29], PM2000 steel [30], AISI 304 stainless steel [31], and IASI 316L stainless steel [32], have also been carried out. The results are promising and some of the problems reported above when using P91/P91 joints seem to tend to be softened, depending on the welding process and conditions used.

Studies in recent years reveal how important it is to know how to properly weld and treat P91 steel welds. Thus, this work aims to deepen the previous studies by performing different PWHT cycles to welded P91 steel samples, analyzing the resulting microstructure, mechanical strength, and hardness.
