Improvement of the Mechanical Characteristics, Hydrogen Crack Resistance and Durability of Turbine Rotor Steels Welded Joints

Abstract

This article is devoted to the following issues: calculating the values of temperatures obtained by simulating welding heating and the subsequent implementation of the welding process at the given mode parameters made it possible to obtain a welded joint of the rotor with an improved initial structure and increased mechanical properties, hydrogen resistance and durability by up to 10–15%; simulating welding heating in the areas of fusion, the overheating and normalization of the HAZ and the formation of austenite grains; specified welding heating creates the conditions for the formation of new products of austenite decomposition in the form of sorbitol in the area of the incomplete recrystallization of the HAZ. In air and gaseous hydrogen, the destruction of the combined joints took place on the weld metal, as well as on the fusion areas, the overheating and the incomplete recrystallization of the HAZ of 20H3NMFA steel as the base metal. Structural materials have a relatively low strength and high fracture toughness in air. This is manifested in a significant reduction in the elongation (δ), the area (ψ) and critical stress intensity factor (K_Ic) of welded joints and the endurance limit of cylindrical smooth rotor steel specimens, which were cut from transverse templates. Welded joints in the whole range of load amplitudes are sensitive to the action of hydrogen.

1. Introduction

Modern requirements for the operation of the United Electric Power System (UEPS) of Ukraine and the Polish Power System (PSE S.A.) (main activity: provide the services of electricity transmission in compliance with the required criteria of the security) with a tendency to major “green” electricity are quite high. The reliability of work and the increase in the power plant turboaggregate (TA) service life are the priority tasks [1]. They are operated in harsh conditions determined by high temperatures and stresses, which lead to irreversible changes in the properties of the metal and damage to the steel weldments [2,3,4,5,6,7,8] (Figure 1 and Figure 2).

Turbogenerator (TG) rotor shafts (Figure 3d) are a complex engineering structure with numerous structural stress concentrators both on the rotor barrel and on its tail. The materials used for the manufacture of such TG assemblies must have sufficient strength and ductility during all TA operating modes [2,3]. So, investigating the heat transfer on the surface and inside TA elements, the stress-strain behavior and influence of hydrogen-containing environments and mechanical loading, the methods of enhancement of cooling media heat transfer, the manufacturing of heat exchangers and other components under mechanical and thermal loading is very useful, because the steam and hydrogen turbine rotors have limited the lifetime of TA.

2. Literature Survey: State of the Art

The steel for shaft and rotor blanks is smelted in open-hearth furnaces with acid lining or electric furnaces. Vacuuming must be used when pouring ingots weighing more than 25 t. According to [2,3,4,5], for the manufacture of shafts and rotors of steam turbines and TG in Ukraine, Poland and other countries, it is recommended to use special grades of steel with the appropriate chemical composition, depending on the strength category. The length of the rotor shafts can reach 13.4 m, and the diameter can reach 1.8 m with a total mass of about 160 t (for the mass of the output ingot over 360 t).

Only the part of the ingot that has the most homogeneous material is used to make the rotor. Magnetic and ultrasonic flaw tests, together with the study of the mechanical properties of the samples obtained by radial and axial drilling and the careful control of the structure, allow for the selection of the necessary material for the rotor. Along with the promising open-hearth method of manufacturing powerful TG rotors, there is the electroslag welding of two workpieces.

The need to develop a new method is associated with the prospect of developing TG for power plants with a capacity of up to 2000 MW. In this case, the diameter of the rotor will approach 3.0 m, and the mass of the rotor shaft can exceed 300 t (respectively, the forging for the manufacture of such a shaft will weigh more than 400 t, and the total weight of the steel ingot is about 680 t).

Therefore, the search for non-traditional methods of manufacturing shafts and rings from electroslag steel for high-speed (3000, 3600 RPM) and low-speed (1500, 1800 RPM) powerful TG and related turbines continues.

At the same time, manufacturers are improving existing technologies for the manufacture and reliable operation of rotors. The parameters that determine the operating conditions of the contact of the wedges with the shaft for TG vary widely. In particular, the nominal bending stress in the shaft varies from 16 … 19 (TVV-320-2) to 29 … 33 MPa (TVV-1200-2), and the average contact pressure varies from 85 (T3V-800-2) to 300 … 305 MPa (TVV-1200-2, current supply groove). The durations of rotor operation before the appearance of developed cracks also differ—from (2 … 3) × 10⁷ cycle (TVV-1200-2) to (2 … 3) × 10¹⁰ cycle (TVB-320-2). For TG type TVV-220-2, the operation duration (5 … 8) × 10¹⁰ cycle [2,3,4] in gaseous hydrogen (used for cooling technology) can be the base for the materials selection for hydrogen turbine rotors.

Low-alloy steels with a typical carbon content of 0.2 … 0.3% are mainly used for steam turbine rotors in EU and the USA. The standard heat treatment of these steels includes austenitization, hardening and tempering at temperatures above 600 °C, which results in a bainite microstructure with a yield strength of 800 MPa [2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34].

The operating time of most of the NPP power units has already exceeded its park resource [25,26,27,28,29,30,31,32,33,34]. Thus, the priority direction of research is to increase the reliability and extend the service life of the operation of power units, including the components of power units.

The increase in operational requirements for the rotors of high-power steam turbines at nuclear power plants necessitates improving the quality characteristics of the initial metal structure of their welded joints, which is characterized by the presence of a certain heterogeneity. It is advisable, by using the optimal welding heating of the manufactured joints, to reduce the level of the initial structural heterogeneity, which increases with an increase in their operating time [25,26,27,28,29,30,31,32,33,34].

In the process of the formation of the initial structure of welded joints, structures can be formed (locally) in it, which can be conditionally referred to as rejection ones. Such rejection structures contribute to the accelerated damage of the metal of the welded joint of the rotor during its operation under fatigue conditions, which causes a decrease in its reliability and a decrease in its resource.

The goal of this work is to simulate the welding heating of the manufactured joint of the steam turbine rotor made of 25H2NMFA and 20H3NMFA steels to obtain the initial structure of the welded joint with an improved quality of mechanical characteristics and crack resistance in hydrogen-containing environments.

3. Formulation of the Problem: Materials and Methods

For the manufacture of forged billets of welded turbine rotors (for example, K-1000-60/1500 and its modifications, K-1000-60/1500-2, etc.), 25H2NMFA steel is used. Blanks for welded rotors are subject to the individual determination of mechanical properties using special quality control methods for their metal.

In the energy sector, the trend of using powerful steam [2,3] and promising hydrogen [4] turbines with a combined rotor, which is obtained by welding, is becoming relevant [1]. This rotor is operated in high- and low-temperature modes. The combined rotor is made of alloy steels that meet the special operating conditions of high-pressure (HP), medium-pressure (IP) and low-pressure (LP) cylinders. For the stages of the rotor, which operates in low temperatures at JSC “Turboatom”, the steels 25H2NMFA and 20H3NMFA were proposed [10,11,12,13,14,15].

For stages operating in the temperature range close to 500 °C, the steel 25H2NMFA was proposed. This steel has relatively high mechanical properties but is difficult to weld.

The welding of the 25H2NMFA and 20H3NMFA steels (

Table 1. Chemical composition of the base metal, electrode wire and weld metal of the combined weld [2,3,4,5,10,34].

Investigated Object	Chemical Composition, Mas %
	C	Si	Mn	Cr	Ni	Mo	V	W	S	P
20H3NMFA	0.20 (0.16–0.24)	0.30 (0.17–0.40)	0.34 (0.25–0.60)	3.0 (2.40–3.30)	0.20 (0.20–0.50)	0.65 (035–0.65)	0.70 (0.60–0.85)	0.46 (0.30–0.50)	≤0.012	≤0.032
25H2NMFA	0.24	0.34	0.46	1.0	2.55	0.46	0.02	-	≤0.014	≤0.024
Wire SV08HN2 GMYu	0.08	0.53	1.05	0.75	2.0	0.45	0.016	-	≤0.012	≤0.018
Weld metal	0.056	0.25	0.98	0.73	1.8	0.45	0.003	-	≤0.012	≤0.018

The content of Co (rotor of the primary circuit of the steam turbine) should not exceed 0.025%, Cu ≤ 0.25.

) is an extremely difficult task [10,11,12,13,14], which is associated with the formation of defective structures in the alloying area, as well as in the area of incomplete recrystallization of the heat-affected zone (HAZ) of the 20H3NMFA steel welded joint. An urgent requirement is also to ensure the necessary resistance against the formation of cold cracks.

The SV08HN2GMYu electrode wire with a diameter of 2.0 mm and AN-43M flux was used for the experiments, which ensured the achievement of the optimal composition in the weld metal of the alloying elements and also limited the presence of harmful impurities of sulfur and phosphorus (Table 1).

The restriction of sulfur and phosphorus prevented the metal from crumbling to weld, which occurs after high tempering.

The welding of the rigid technological samples was performed on the mode: I_weld = 300 … 320 A; U_weld = 34 … 36 V; V_weld = 18 m/h. Control over the formation and growth of cracks was performed using acoustic emission signals. It was established that cold cracks are formed in the weld metal during the welding of technological samples without preheating. Their origin, as shown by acoustic emission signals, begins immediately after welding and continues with the imposition of subsequent rollers.

The resistance to failure of 20H3NMFA welds in unheated welding was relatively low, but when heated to 300 °C, such indicators become satisfactory. Thus, to ensure a high resistance to the slow destruction of welded joints of the steels 25H2NMFA and 20H3NMFA, obtained by automatic submerged arc welding, it is possible to recommend the pre-heating and concomitant heating of samples in the temperature range of 250 … 300 °C.

The proposed technology, as shown by metallographic studies (Figure 4), allowed for the obtention of welded joints with relatively high mechanical properties, without cracks and tears.

The use of wire, as well as the above modes of welding, provided the structure of the weld metal in the form of bainite with carbides. This structure is not prone to slow destruction, which is ensured by the absence of the local microplastic deformation of its structural components.

The determination of the mechanical properties was performed on samples cut from weld metal, as well as from combined welded joints of the steels 25H2NMFA and 20H3NMFA, the thickness of which was 80 mm. Welding was performed according to the above modes. Welded joints were subjected to high tempering at 620 °C for 30 h.

It was found that the strength and ductility of the welded joints at 20 °C and 450 °C (operating temperature) are relatively high and meet the regulatory requirements (

Table 2. The toughness of the weld metal of the combined welds after tempering at 630 °C for 30 h.

Investigated Object	Impact Toughness KCV, J/cm2
	20 °C	−20 °C	−40 °C
20H3NMFA	105 … 78 91.5	60 … 45 55	40 … 30 35
25H2NMFA	136 … 120 128	81 … 40 60.5	60 … 30 45
Weld metal (wire SV08HN2GMYu)	100 … 72 86	50 … 35 42.5	39 … 31 35

). The toughness of the weld metal is slightly lower than that of the base metal, which was not affected by welding heat (Table 2 and

Table 3. Impact toughness of samples with a V-notch along the HAZ (fusion and overheating areas) (Figure 4).

Investigated Object	Impact Toughness KCV, J/cm2
	Suggested Welding Process	Standard Welding Process
Sample 1 (fusion area)	74	64
Sample 2 (fusion area)	59	57
Sample 3 (fusion area)	69	60
Sample 1 (overheating area)	187	176
Sample 2 (overheating area)	171	168
Sample 3 (overheating area)	88	178

). The critical temperature of fragility is in the region of low temperatures (−15 °C … −10 °C).

The metallographic analysis of the structure of combined welded joints showed that the metal of HAZ sections has a mainly sorbitol and bainitic-ferritic structure on the 25H2NMFA side, as well as a bainitic martensite structure on the 20H3NMFA side [10,11,12,13,20].

The weld metal is characterized by the presence of a bainitic-ferritic structure. Measurements of the hardness of the metal of the combined weld showed that the structures in the area of overheating on the side of the 20H3NMFA steel have a hardness of HV 350 … 360, and in the area of overheating of the 25H2NMFA steel, they have a hardness of HV 300 … 320.

The base metal, which has not undergone welding heat, has a hardness of HV 280 … 290 (steel 20H3NMFA) and HV 230 … 240 (steel 25H2NMFA). After tempering at 620 °C (20 h) the structure of the welded joints acquires alignment, and the hardness is HV 180 … 190.

In the area of the fusion of the weld metal with the steel 25H2NMFA, the hardness decreases to HV 160 … 170. In the area of the fusion of the weld metal with the steel 20H3NMFA, the hardness is slightly higher than HB 190 … 205.

The hydrogen content in the metal of the ingots should not exceed 2.23 ppm [10,11,12,13,14,15], and the decrease in its value in steel and weldments has led to the improvement of the mechanical characteristics, hydrogen resistance and durability of steam turbine rotor steels welded joints. The initial steel structure is tempered bainite (Figure 4).

The study of the structural state of welded joints made of 25H2NMFA steel, as well as of steels with a similar chemical composition [6,10,11,12,13,14], showed that rejection structures or structures close to rejection ones can form in the metal of welded joints. For example, it is still an urgent task to obtain fine austenite grains in the areas of the heat-affected zone (HAZ) of welded joints: fusion, overheating and normalization. A significant contribution to the practical and theoretical solution of the problem of obtaining a fine-grained austenitic structure was made in [6,10]. However, his work could not lead to a solution to the problem of preventing the formation of large austenite grains in thick-walled welded joints.

Welding heating (standard technology) provides a long stay of fusion, overheating and normalization of the HAZ of the studied thick-walled welded joints in the temperature range Ts-950 °C, which leads, accordingly, to the formation of large austenite grains. Long-term heating to the temperature range A_C1–A_C3 leads, in the area of the incomplete recrystallization of the HAZ, to the formation of new products of austenite decomposition in the form of globular pearlite [11,12,13,14]. In the samples, after welding, a significant increase in hardness was revealed, corresponding to the areas of fusion and overheating of the HAZ [11,12,13,14]; however, they did not explain the relationship between the increased hardness and the structural state of the areas.

The presence of the above structures reduces the resistance of the metal of welded joints to brittle fracture from the action of centrifugal forces, stress corrosion, hydrogen cracking and fatigue damage under the action of alternating stresses during rotation, bending and torsional vibrations.

The improved thermal problem [11,12,13,14], applied to the welded joint of the rotor (Figure 5), was solved in a joint delivery under the conditions of the Navier–Stokes (molten metal of the weld pool) and Fourier (base metal, as a solid phase) laws (Figure 6).

The solution was carried out in a cylindrical coordinate system in an axisymmetric setting. It was assumed that a quasi-stationary process of heat transfer and crystallization takes place.

Let us write the Navier–Stokes Equation (1):

$$\{\begin{array}{l}\frac{\partial \xi }{\partial t}+\frac{1}{r}\frac{\partial \psi }{\partial r}\frac{\partial \xi }{\partial z}-\frac{1}{r}\frac{\partial \psi }{\partial z}\frac{\partial \xi }{\partial r}+\frac{1}{r}\frac{\partial \psi }{\partial z}\frac{\partial \xi }{r}=\\ =V\left[\frac{1}{r}\frac{\partial }{\partial r}\left(\frac{\partial \xi }{\partial r}\right)+\frac{{\partial }^2\xi }{\partial z^2}-\frac{{\xi }^2}{r}\right]+g\beta \frac{\partial T}{\partial r}+\overrightarrow{\nabla }\prime \cdot \left(\overrightarrow{j}\times \overrightarrow{B}\right)\\ \frac{\partial }{\partial z}\left(\frac{1}{r}\frac{\partial \psi }{\partial z}\right)+\frac{\partial }{\partial r}\left(\frac{1}{r}\frac{\partial \psi }{\partial r}\right)+\xi =0\\ \frac{\partial T}{\partial t}-\frac{1}{r}\frac{\partial \psi }{\partial z}\frac{\partial T}{\partial r}+\frac{1}{r}\frac{\partial \psi }{\partial r}\frac{\partial T}{\partial z}z=\\ =\frac{1}{\rho c}\left[\frac{1}{r}\frac{\partial }{\partial r}\left(kr\frac{\partial T}{\partial r}\right)+\frac{\partial }{\partial z}\left(k\frac{\partial T}{\partial z}\right)+Q_k\right]\end{array}$$

(1)

Here, υ = μ/ρ is the kinematic viscosity; g—acceleration of gravity; vector $\overrightarrow{\nabla }\prime =-\frac{\partial }{\partial r}{\overrightarrow{e}}_z+\frac{\partial }{\partial z}{\overrightarrow{e}}_r$. The mixed product $\overrightarrow{\nabla }\prime \cdot \left(\overrightarrow{j}\times \overrightarrow{B}\right)$ determines the influence of electromagnetic forces on the behavior of the dynamics of flows of liquid metal in the molten bath; $\overrightarrow{B}$ is the magnetic field strength; β is the coefficient of thermal expansion; T is the temperature; ψ is the melt flow function; ξ—coordinate of the rotor of the velocity field; ρ, s, μ, k—density, heat capacity, dynamic viscosity and thermal conductivity; Q_k—additional heat input into the bath melt.

Let us introduce the stream function of the molten metal ψ and the vortex ξ. The relationship with physical variables will be as follows:

$$\xi =\frac{\partial \nu }{\partial z}-\frac{\partial u}{\partial r};u=\frac{1}{r}\frac{\partial \psi }{\partial r};V=-\frac{1}{r}\frac{\partial \psi }{\partial z}$$

(2)

According to the approximation, System (1) is written in the Form (3):

$$\{\begin{array}{l}{\displaystyle \underset{\Omega }{\int }\left[\frac{\partial \psi }{\partial r}\frac{\partial \xi }{\partial z}-\frac{\partial \psi }{\partial z}\frac{\partial \xi }{\partial r}+\frac{\partial \psi }{\partial z}\frac{\xi }{r}\right]}{\phi }_{k}drdz+\\ +\nu {\displaystyle \underset{\Omega }{\int }\left[(\frac{\partial \xi }{\partial r}{\phi }_k+r\frac{\partial {\phi }_k}{\partial r}+\xi \frac{\partial {\phi }_k}{\partial r}+\frac{\partial \xi }{\partial z}\frac{\partial {\phi }_k}{\partial r}r)\right]}{\phi }_{k}drdz+\\ +gB{\displaystyle \underset{\Omega }{\int }T({\phi }_k+r\frac{\partial {\phi }_k}{\partial r}){\phi }_{k}drdz={\displaystyle \underset{\Omega }{\int }\pi (\overrightarrow{j}\times \overrightarrow{B)}{\phi }_{k}drdz}}\\ {\displaystyle \underset{\Omega }{\int }\left[\frac{\partial \psi }{\partial z}\frac{\partial {\phi }_k}{\partial z}+\frac{1}{r}\frac{\partial {\phi }_k}{\partial r}({\phi }_k+r\frac{\partial {\phi }_k}{\partial r})+\xi {\phi }_{k}r\right]}{\phi }_{k}drdz=0\\ {\displaystyle \underset{\Omega }{\int }\left[(\frac{\partial \psi }{\partial z}\frac{\partial T}{\partial r}+\frac{\partial \psi }{\partial r}\frac{\partial T}{\partial x})+\frac{k}{\rho C_p}(\frac{\partial T}{\partial r}\frac{\partial {\phi }_k}{\partial r}+\frac{\partial T}{\partial z}\frac{\partial {\phi }_k}{\partial z})\right]}{\phi }_{k}drdz=\\ ={\displaystyle \underset{\Omega }{\int }{Q}_k}{\phi }_{k}drdz,(k=1,\dots ,n)\end{array}$$

(3)

φ_k are basis functions, which were determined by the method of R-functions.

The free surface of the molten bath Ω₃ (Figure 6) was taken as flat. With regard to the weld pool, a joint solution of the nonlinear system of equations of the magnetohydrodynamics of the motion of molten metal as a viscous liquid, the system of Maxwell’s equations for the distribution of the vectors of strength of the electric and magnetic components of electromagnetic fields, was performed. With regard to the base metal, as a solid phase, to determine the values of temperatures T, the nonlinear equation of thermal conductivity was solved. System (3) was formulated as a problem as applied to phase transitions with a free boundary, which corresponds to the conditions of the Stefan problem.

In the calculations of the temperature regime in welded joints, the traditional scheme (which coincides with the Ritz method) was used; in this case, the basis functions were constructed in spatial variables, and the motion in time was taken into account using the Rothe method, i.e., by sampling in time. This method made it possible to reduce the system of non-stationary Navier–Stokes equations to a system of stationary equations. The solution of the last system was carried out in an iterative way, in which the k + 1 approximation includes the k-th approximation as the initial one.

By choosing a system of basis functions {φ} in the form of splines (the classical method), we projected the resulting expansion of solutions into an n-dimensional subspace. In this case, the time for calculating all the integrals that are included in the matrix of the system of linear Navier–Stokes equations is significantly reduced. With the optimal selection of the initial basis functions (taking into account the experimental data), fairly accurate results were obtained already on the first few terms of the expansion.

The solution to the heat problem allowed:

1. For the establishment of the temperature regime of the welding process, ensuring the formation of a given structure of the welded joint;

2. For the revelation, in the weld metal and in the area of HAZ fusion, of the places of local welding overheating, where structures can form as rejection ones or as structures that can be referred to as rejection ones. For example, large austenite grains in the areas of fusion, overheating and normalization of the HAZ and new products of austenite decomposition in the form of globular pearlite in the area of incomplete recrystallization. It was found that the reduced structures can form mainly in the central zone of the welded joint (see Figure 5 and Figure 7) (Samples 1–3). Note that the simulation of welding heating allows, in a practical way, for the prevention of the formation of both rejection structures and structures that can be attributed to rejection [6].

The use of numerical data characterizing the temperature regime of welded joints made it possible, taking into account the known techniques [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35], to optimize the parameters of the automatic welding regime. Then, the welding process of the prototype witness was carried out on the optimized parameters: welding current 390–420 A; arc voltage 38–40 V; welding speed 20–25 m/h; electrode wire feed speed 125–130 m/h; diameter of the electrode wire 2.0–2.5 mm.

The mode parameters during the welding process were changed within the recommended values. Their change was caused by the need to obtain an optimized temperature regime that ensures the formation of a given structure.

The preliminary and concomitant heating of the prototype to be welded was 300–350 °C. Immediately after welding, the sample was subjected to high tempering T = 630–650 °C, lasting 130–150 h. Then, from the witness sample (Figure 5), templates were cut out to study the structure and properties (Figure 7).

4. Results and Discussion: Weldments Structure and their Mechanical Properties

The application of the famous method, when solving the heat problem, made it possible to construct smoothly-approximated temperature isotherms (Figure 8 and Figure 9), which made it possible to restrict the HAZ sections with the same structure in the welded joint.

The calculated and experimental data characterizing the regions of the formation of the corresponding structures were compared with the data of the thermos-kinetic diagram of the steels 25H2NMFA and 20H3NMFA. The comparison, as well as taking into account the known results [22,23,24,25,26,27,28,29,30,31], made it possible to improve the mathematical model of welding heating of the manufactured rotor and to refine the numerical data characterizing the thermal cycle.

The widths of the HAZ sections on the templates (Figure 4) were determined by taking into account the presence of similar structures in the areas. The separation of differing structures in relation to the corresponding sections of the HAZ was marked with reference points. The study of the structural state of the welded joint made it possible to reveal the width of the HAZ sections (Figure 5).

Template 1: fusion area 0.1–0.12 mm; overheating area 3.0–3.1 mm; area of incomplete recrystallization 2.4–2.6 mm. Template 2: fusion area 0.11–0.13 mm; overheating area 3.2–3.3 mm; area of incomplete recrystallization 2.6–2.8 mm. Template 3: fusion area 0.1–0.11 mm; overheating area 3.1–3.2 mm; area of incomplete recrystallization 2.5–2.7 mm.

Seam metal (Figure 4 and Figure 6, Template 1) has a bainitic structure with sorbitic components oriented in accordance with the temperature regime providing their directional formation.

On template 2, the structure is presented as bainite-troostite with sorbite inclusions with a locally thickened character. On template 3, the structure is also bainite-troostite with a small amount (about 8%) of the ferrite component.

In the fusion area of HAZ (Figure 10), there is a smooth transition between the structure of the weld metal and the base metal. The structure of the fusion area is characterized by the presence of fine grains in a dark matrix. There are precipitates of the cementite type in the form of rounded, finely dispersed inclusions, the arrangement of which along the body and along the grain boundaries of the α-phase is close to uniform. The structural heterogeneity of the fusion area meets the regulatory requirements.

The overheating section (Figure 11 (template 2, Figure 7 and Figure 8)) has a predominantly sorbate-troostite structure.

Austenitic grains in the areas of fusion and overheating correspond to No. 7–No. 9. Accordingly, the largest grains are noted on template 2 (see Figure 11). Smaller ones are noted on templates 1 and 3. It was revealed that the matrix phase in the areas of fusion and overheating of the HAZ is bainite, which is close to granular in shape. The formation of bainite in the process of post-weld cooling occurs mainly by the martensitic mechanism.

The structure of the area of incomplete recrystallization of the HAZ (Figure 12) represents temper bainite with sorbitol constituents (dark grains in bainite).

Sorbitol constituents are the new decomposition products of austenite. It was found that M₃C carbides in the structure of the HAZ sections and in the weld metal structure are distributed mainly evenly. The grain size of the α-phase under tempering conditions does not undergo changes. The coagulation of carbides of the first group (M₇C₂₃, M₂₃C₆) is insignificant, and in the second group, Mo₂C, it is absent.

It was found that the amount of retained austenite (7–9%) in the weld metal, which turns into a ferrite-carbide mixture during tempering, does not significantly affect the mechanical properties, which is confirmed by the microhardness values (Figure 13).

The presence of the considered structures in the metal of the welded joint is also confirmed by the value of the microhardness.

It was found that the specified welding heating ensures the formation of smaller austenite grains in the HAZ sections, which is confirmed by the values of the impact toughness increasing by 10–15% in comparison with the similar values of the samples manufactured using the standard technology.

It was found that the average short-term mechanical properties of welded joints (see Figure 7) exceed the regulatory requirements by 5–10%: σ_B 680 N/mm²; σ_0.2 520 N/mm²; δ 25%; ψ 63%; KCV 187 J/cm²; HB 197.

The study of the welded joints’ structure has shown that high tempering in the weld metal provides a transition from α-phase crystals to chromium and molybdenum carbides, as well as the formation of new carbides: M₇C₂₃, M₂₃C₆ and Mo₂C. The value of the critical point A_C1 decreases by 32–35 °C in comparison with the base metal that does not undergo welding heating. The value of the temperature of the A_C3 point of the base metal, in comparison with similar values, does not change noticeably. The density of dislocations in the crystals of the α-phase after tempering decreases by about 10–15% and amounts to 2.15 × 10⁹ cm⁻² in the fusion area, 1.9 × 10⁹ cm⁻² in the overheating area and 1.7 × 10⁹ cm⁻² in the base metal.

5. Weldments’ Static Crack Resistance in Hydrogen

Compact specimens with welds, which provide controlled propagation of the main crack on the metal of different zones of the weld (Figure 14 and Figure 15) [35,36,37,38,39,40], effectively contribute to the optimization of welding modes.

The disadvantage of the model in the case of welded joints of different configurations is that, in the process of spreading cracks along the entire length of the part, welds can serve as traps for cracks, and this changes their trajectory (Figure 16), sometimes at an angle up to 90° (Figure 16). This phenomenon is due to the occurrence of stress concentrators in the remelting areas, which negatively affect the performance characteristics, particularly crack resistance. During welding, microcracks appear near the fusion line, which cause a decrease in compressive residual macro stresses. All of this also negatively affects the performance of machine parts under cyclic and static loads during long-term operation.

Uneven grain sizes, the sizes and distribution of reinforcing phases, local thermal stresses and other defects cause the significant sensitivity of various structural elements of welded joints to the action of hydrogen [41,42,43,44,45,46]. Even the simulation of the soldering regime with the short-term (15 min) heating of KhN43MBTYu and 05Cr19Ni55 alloys to 1473 K leads to an increase in grain size and the concentration of large grain boundary intermetallic precipitates, which significantly enhances the hydrogen embrittlement of materials [47,48]. In the presence of hydrogen, the mechanical characteristics of welded joints deteriorate—short-term strength and ductility [43,44] and low-cycle durability [44,45].

Important characteristics of critical structures, including steam and hydrogen turbine rotors, are the parameters of crack resistance [49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118], which are also significantly reduced by hydrogen [41,42,46,47,48,53,54,55,56] with a pressure up to 10 MPa. Given the variety of factors that determine the properties of welded elements in hydrogen-containing environments, to assess their performance requires the experimental determination of a set of physical and mechanical properties of a particular joint [119,120,121,122,123,124,125], especially for the structural elements of the hydrogen energy buffer (electrolizers, fuel cells, hydrogen storage and grid distribution), with the intention of utilizing hydrogen and the accompanied phenomenon of their hydrogen degradation during long-term service.

Four types of samples were tested: smooth five-fold cylindrical with a working part diameter of 5 mm to determine the short-term strength and ductility and the fatigue life; flat with a rectangular cross-section of 3 × 6 mm and a length of the working part of 20 mm for the study of low-cycle fatigue; 25 mm-thick rectangular compact specimens with an off-center tensile speed of 0.1 mm/min to assess fracture toughness. The critical values of SIF in air were determined by the J-integral method [55], because the plastic characteristics of the alloy in an inert medium are high (

Table 4. Mechanical properties of welded joints in air (upper value) and in hydrogen at a pressure of 10 MPa (down value).

Investigated Object	T, °C	σ0.2, MPa	σB, MPa	δ, %	ψ, %	KIc, MPa∙m 1/2
Steel 20H3NMFA	20	620 610	730 710	11 8	32 21	112 61
	450	510 510	600 610	8 7	27 25	92 83
Steel 25H2NMFA	20	520 530	687 680	14 10	40 26	118 72
	450	460 430	570 580	12 12	40 36	101 91
Welded joint	20	530 510	690 690	24 11	67 31	39 21
	450	450 460	570 560	16 15	63 52	42 38

), and at a specimen thickness of 20 mm, the plane strain state (PSS) is not realized.

The fracture toughness under elastic-plastic fracture was estimated by the J-integral method using the dependence K²_Ic(J) = J_IcE/(1 − μ²), where E is the modulus of elasticity (Young’s modulus) and μ is the Poisson’s ratio [51]. All roofs and fatigue cracks are located in the middle of the weld metal (Figure 8, Section 5). In the initial state after heat treatment, the structure of the steels 25H2NMFA and 20H3NMFA consists mainly of bainite tempering and a small amount of ferrite-carbide mixture.

Such materials have a relatively low strength and high viscosity of fracture in air (Table 4). Under the hydrogen action, the strength characteristics of hydrogen do not change. At short-term stretching at room temperature, its effect is manifested in a significant reduction in the relative elongation δ, the relative transverse narrowing ψ and the critical stress intensity factor (K_Ic) of steels and, especially, welded joints. The viscosity of the fracture, which is characteristic of many welds, deteriorates significantly (by 40–50%) [41,42].

For example, the reduction in the fracture toughness in the base metal X80 steel increased with increasing current density; the difference in the effects of hydrogen on the fracture toughness of the base metal and the weld was attributed to the specific microstructural features of both materials [41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90]. Compared with the weld, the base metal exhibited a more refined microstructure. The higher fraction and the larger grain boundary density were conducive to crack arrest performance in gaseous hydrogen-containing environments [41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90].

In comparison to similar H-free samples, the H-charged samples presented lower fracture toughness. For samples, with notches located at the base metal, stir zone and heat-affected zone metal, the average Critical Crack Tip Opening Displacement decreased from 0.96 to 0.25 mm, from 0.48 to 0.43 mm and from 0.22 to 0.08 mm, respectively [42]. At the working temperature (450 °C), the effect of hydrogen is negligible (Table 4).

Experiments have shown that, in air and hydrogen, the destruction of the combined joints took place on the weld metal, as well as on the fusion areas, the overheating and the incomplete recrystallization of the HAZ of 20H3NMFA steel as the base metal.

6. Hydrogen Influence on the Welded Joint Durability under Cyclic Loadings

The operation of many structures—particularly, steam and hydrogen turbine rotors, cooling infrastructures of powerful semiconductor devices and industrial single-crystal growth applications [3,4,5,6,7,8,9,25,26,27,28,29,30,31,32,33,34,45,55,56,57,58,59,79,80,81,82,83,84,85,86,87,88,89,90,91,110]—is accompanied by their cyclic loads, which often exceed the yield strength of the material [52,54,55,56], based on understanding the hydrogen embrittlement of materials from the atomistic level to the continuum [100,101,102,103,123,124,125,126,127,128]. In such cases, low-cycle fatigue tests most fully reflect the operating conditions of the products. Rigid low-cycle bending is also used to assess the water resistance of steels and alloys due to the simplicity of the practical implementation and combination in the near-surface layers of samples of maximum stresses and concentrations of hydrogen [57,58,59].

Low-cycle endurance was investigated by the rigid zero-zero pure bending of flat samples with the dimensions of the working part (3 × 6 × 20 mm) in the range of deformation amplitudes of 0.3 … 1.25%. From such amplitudes, the load frequency of 0.83 Hz ensured the absence of heating of the samples and the short duration of the experiments and allowed for the detection of the effect of hydrogen on the number of cycles before failure [57,58,59].

It was found that, in the studied range of amplitudes in the logarithmic coordinates of the load amplitude, the number of cycles to failure is represented by straight lines in both air and hydrogen, as described by the Coffin–Manson equations (Figure 17). It is known [57,58,59] that, under low-cycle loading, the magnitude of the deformation amplitude ε determines the time to crack formation, the stress-strain state at the crack tip and, when tested in the presence of hydrogen, the amount and nature of the distribution in the sample.

In steels with a high viscosity and a low yield strength, the level of local micro stresses for crack germination requires a high concentration of hydrogen [119,120,121,122,123,124,125], which decreases with the increasing frequency and amplitude of the load, which probably causes the weakening of the hydrogen embrittlement steels 25H2NMFA and 20H3NMFA (Figure 18, curves 1, 2).

The opposite pattern was found in tests of 17G1SU steel [54]—for large amplitudes of cyclic tensile, the effect of the corrosive environment on the number of cycles to the destruction of the parent metal and the weld is greater.

20H3NMFA steel with a higher strength and lower ductility has a lower durability and is stronger in hydrogen than 25H2NMFA steel. In the whole range of amplitudes, the influence of hydrogen on welded joints is very strong (Figure 18) and is illustrated by β_N—the coefficient of hydrogen influence on the number of cycles to failure of samples made of 25H2NMFA and 20H3NMFA steels and their welded joint during low-cycle bending (β_N = N^H/N^He).

The endurance limit of cylindrical smooth specimens that rotate was determined by their bending. Cylindrical specimens for fatigue tests were cut from transverse templates of butt-welded joints. In the whole range of load amplitudes, the number of cycles before the destruction of combined welded joints is very sensitive to the action of hydrogen (Figure 19) and is significantly less than that of the rotor steel 38KhN3MFA [5].

An analysis of the improvement of up to 10–15% of the mechanical characteristics, hydrogen crack resistance and durability of rotor steels welded joint is possible in order to obtain an initial structure by the overheating and normalization of the HAZ and the formation of austenite grains reduced in size. This revealed that the specified welding heating creates conditions for the formation of new products of austenite decomposition in the form of sorbitol in the area of incomplete recrystallization.

7. Conclusions

It was established that the calculated values of temperatures obtained by simulating welding heating and the subsequent implementation of the welding process at the given mode parameters made it possible to obtain a welded joint of the rotor with an improved initial structure and increased the mechanical properties, hydrogen resistance and durability by 10–15%.

By simulating welding heating in the areas of the fusion, overheating and normalization of the HAZ, the formation of austenite grains reduced in size was provided.

It was revealed that the specified welding heating creates conditions for the formation of new products of austenite decomposition in the form of sorbitol in the area of the incomplete recrystallization of the HAZ.

It was shown that, in air and hydrogen, the destruction of the combined joints took place on the weld metal, as well as on the fusion areas, the overheating and the incomplete recrystallization of the HAZ of 20H3NMFA steel as the base metal. Structural materials have a relatively low strength and high viscosity of fracture in air.

At short-term stretching at room temperature, its effect is manifested in a significant reduction in the relative elongation δ, the relative transverse narrowing ψ and the critical stress intensity factor (K_Ic) of steels and, especially, welded joints. The fracture toughness, which is characteristic of many welds, deteriorates significantly (by 40–50%).

The endurance limit of cylindrical smooth specimens of rotor steel which were cut from transverse templates of butt-welded joints in the whole range of load amplitudes and the number of cycles before the destruction of combined welded joints are very sensitive to the action of hydrogen.