The Use of the Linear Energy Calculation Model in High-Frequency Induction (HFI) Tube Welding Technology to Obtain Optimal Microstructure and Weld Geometry

Abstract

The article presents a calculation model of the linear energy of welding P235GH steel tubes with high-frequency currents in order to obtain an optimal microstructure and geometry of the weld of high internal purity. The model was developed based on real data for the standard linear energy used in the steelworks Huta Łabędy and presented as the power factor P/V and P/(V·t), where P is the power [kW], V the production speed [m/min] and t the wall thickness. The model can be used for two ranges of pipe diameters: 114.3–168.3 mm and 219.1–323.9 mm. The data from the model were implemented into the High Frequency Induction (HFI) control panel of Huta Łabędy in order to produce test tubes which were subsequently tested with ultrasounds to verify the quality of the internal weld. In addition, samples were taken for metallographic analysis, which was supposed to check whether the applied linear energy calculation model allows the obtainment of the optimal weld geometry and the optimal angle of the metal flow line allowing for swelling and the extrusion of melted impurities from the inside of the joint by the squeeze rolls. The metallographic analysis also determined the nature of the occurrence of ferrite inside of the center diffusion bond and the zonal microstructure of the joint, the control of which is based on the correlation of the parameters of the mechanical process of forming the tube with the linear energy of welding. Carrying out the technological and technical process based on the applied HFI linear energy calculation model allowed us to obtain a weld of high purity and metallurgical consistency. This model can be used in the future on an industrial scale for the production of pipes using the HFI method.

1. Introduction

The technology of steel tube production by joining inductively heated edges of sheet metal is extremely important from the industrial point of view. The very process of the production of longitudinally welded tubes has been known and used for a long time based on various technologies evolving over time [1]. These technologies are now widely used in industry due to their properties and advantages: high energy efficiency, excellent product quality, very good accuracy in heating specific zones in a short time and clean working conditions. Thanks to this, HFI heating is used in a number of processes: induction hardening, tempering, soldering, welding, annealing, preheating and postheating, forging, melting, straightening, etc. [2]. In addition, HFI processes can be successfully used for both non-alloy and alloy steels and for copper, brass, aluminum, titanium and gold [3] although other techniques can be applied [4,5].

The HFI induction heating is most often performed by high-frequency heating because the efficiency of this process is the highest and, at the same time, it is possible to influence the depth and intensity of the heating process to some extent [6]. This heating frequency is characterized by a very high energy density and thanks to this, it is possible to obtain local melting of the material before the heat from the heating area spreads throughout the entire volume of the material (induction-heated materials usually have high coefficients of thermal conductivity which promotes the spread of heat) [7]. Using the properties of the process described above, it is possible to cross-heat the material and heat the surface for the purpose of surface hardening and melting metals. The latter property was used in the process under analysis. Intensive induction heating leads to local melting of the metal surface. When carbon steel tubes are welded, an electrical voltage is induced across the edge of the open tube immediately before closing it, in a narrow gap. This voltage causes the current induced along the edges to flow to the point of the edge connection which causes the metal to heat up rapidly. The force exerted by the pressure rolls causes the molten metal to join on both sides, creating a hot diffusion bond. At the same time, the squeezing force pushes impurities out of the welded cross-section, resulting in a weld structure similar to the microstructure obtained as a result of draft instead of the typical casting structure which arises in most welding processes [8].

The development of induction heating technology, successfully used in an increasingly wider area of applications, is also facilitated by the development of semiconductor converter technologies operating with ever higher efficiencies and ever wider ranges of currents and voltages [9].

The latest technological trend in this area is resistance welding using high-frequency currents through induction heating for ever larger diameters, up to 660 mm, and wall thicknesses, up to 25 mm, which shows that the HFI technology has a clear potential to provide the energy market with cheaper options for the production of thick-walled tubes [10].

Improvements in the HFI induction welding process are also moving toward the development of current generators and toward the formation of an optimal distribution of the electromagnetic field. The development of HFI current generators is predominantly related to the technological development of components used in power generators. On the other hand, the development of the field distribution method is based on the development of calculation methods and the improvement of the impeder material based on experimental research on welding equipment [11]. Also, our research focused on the use of calculation methods for linear energy used in HFI welding and guaranteeing the highest quality, as determined by the optimal geometry of the weld microstructure and swelling lines, i.e., the metal flow line.

It should be remembered that HFI technology depends on the steel grade. In the case of ordinary non-alloy carbon steels, the optimization of welding energy and thus obtaining control over the weld in its individual zones is associated with the correlation of the main parameters monitored and controlled during the high-frequency welding process of tubes (welding speed V [m/min] and supply power P [kW]) and adapting them to the dimensions of the tube (thickness and diameter [mm]). The situation is different for other types of steel grades. For example, for the High-strength low-alloy steel (HSLA), the weldability is also affected by the width of the strip used for the production, the quality of the strip edges, the size and uniformity of extrusion of the impurities from the welding point and finally the optimization of the heat treatment after welding [7].

Also, the very microstructure of the HFI tube weld depends on the type of steel being welded, which is why it is also described differently by scientists and it is important to determine it properly as well as standardize the nomenclature and microstructure of individual weld zones, in particular regarding the center line often referred to as the “decarburization line”, “ferritic line” or “bond line” [12].

To conclude, HFI technology is successfully used by many pipe manufacturers but still requires process improvements, in particular in the management of heat distribution and control of the shape of the weld with metal flow lines [13]. Previous geometry studies focused on the influence of the quality of the edges of the welded strip on the size of the Heat affected zone (HAZ) [14]. Also, the microstructure of the HFI joint was variously classified and named by the authors and particular controversies aroused the terms concerning the center line of the joint. There are also no studies on the shape and dimensions of the pipe joint in correlation with the technological parameters of the HFI welding process. What is worth emphasizing is that there are no HFI welding simulators in the industry that would help pipe manufacturers in selecting technological parameters. Therefore, the research presented in the publication is a response to these needs and focuses on a mathematical model of the HFI welding process that can be used on an industrial scale. The research also presents a description of the geometry and dimensions of the joint with a characteristic hourglass shape related to the linear energy and pipe wall thickness and dispels doubts about the microstructure of the joint.

The publication presents the use of the computational model to improve the HFI welding technology in order to increase the efficiency of the process.

The publication aims to prove that the ability to control the weld in the technological process based on a computational model optimizing the amount of linear energy of welding is a decisive element affecting the quality and scale of the reduction in production costs with optimized heat supply [15].

2. Materials and Methods

The methodology used in the research on the calculation model of the HFI tube welding was based on data obtained from the tube production line in the steelworks Huta Łabędy, Gliwice, Poland. The presented microstructural analysis was carried out in the laboratory of the Częstochowa University of Technology, Poland.

2.1. Description of the Production Line Used in the Experimental Research

The experimental research was carried out in the induction heating system of Huta Łabędy used for longitudinal HFI welding of tubes and the idea is presented schematically in Figure 1. The manufacturer of the line is SMS Meer, Monchengladbach, Germany.

The heating occurs due to the flow of eddy current within the structure of the tube. The area of current flow is marked with dashed lines. These are the source currents generated in the inductor and the eddy currents generated in the tube material. To ensure that the density of eddy currents in the welding area is sufficient, we need to concentrate the magnetic field at the welding point. For this purpose, an element concentrating the magnetic field in the welding area is used in the system, commonly called an impeder [16]. It is a mandrel made of ferromagnetic material placed inside the tube.

The strip is shaped in each stage by the forming rolls: it firstly moves into an open tube which passes through a high-frequency inductor and as a result of this, a high-frequency ring current is induced in the tube causing the edges of the sheet to heat up and join at the welding point [17]. The temperature required for the welding is generated by resistive heating of a narrow zone along the edge of the strip. The heated edges of the strip are pressed against each other by squeeze rolls; this produces a homogeneous longitudinal weld without additional binder material and with a granular structure (as after forging), unlike conventional welds which have a casting structure [18]. The flash of metal with impurities arising on the inner and outer surfaces of the tube due to the pressure of the heated edges is cut flush with the level of the tube surface using special knives made of cemented carbide [6]. Immediately afterward, the HFI weld is subjected to a 3-step induction annealing treatment to ensure that the properties of the microstructure in the weld area match those of the base material [9].

The HFI welding parameters are correlated with the two impeders in the diameter ranges from 114.3 to 168.3 mm (circle-shaped impeder filled with ferrites) and for diameters from 219.1 to 323.9 mm (horseshoe-shaped impeder). The length of the ferrites for both diameter ranges is identical: 600 mm. The end of the impeder is raised towards the welding point to the distance of the thickness of the welded tube wall. The inductors are selected depending on the diameter of the tube, taking into account the internal clearance (i.e., the distance between the inner diameter of the inductor and the outer diameter of the tube, which is about 20%) and the length of the inductor (equal to the diameter of the tube). The inductor is always positioned as close as possible to the welding point, as allowed by design constraints and space availability, i.e., at a distance of about 20 to 30 mm from the side rolls of the squeeze welding cage.

Figure 2, Figure 3, Figure 4 and Figure 5 show the welding station used during the tests. This station consists of an inductor, an impeder and a squeeze cage comprising a bottom roll, two side pressure rolls and two oblique rolls holding the edges of the welded sheet in one plane.

2.2. HFI Welding Calculation Model

The calculation model was developed based on experimental data from production campaigns, including the welding of tubes with diameters from 114 mm to 323 mm and wall thicknesses t from 3.6 mm to 5.6 mm, which were recorded in the IBA analyzer 6.10.5 software for which the best weld quality was obtained (

Table 1. Actual data of the HFI welding process.

Pipe Dimensions		Process Parameters
Diameter [mm]	Thickness, t [mm]	Power, P [kW]	Speed, V [m/min]	P/V [kW·min/m]	P/(V·t) [kW·min/m·mm]
114	3.6	258	35	7.4	2.0
114	3.6	273	38	7.2	2.0
139	3.6	280	38	7.4	2.0
139	3.6	293	40	7.3	2.0
139	3.6	268	36	7.5	2.1
139	4	308	38	8.1	2.0
139	4	277	34	8.1	2.0
139	4	239	30	8.0	2.0
168	4	304	38	7.9	2.0
168	4	299	36	8.3	2.1
168	4	270	30	9.0	2.2
168	4.5	333	34	9.8	2.2
168	4.5	303	30	10.0	2.2
168	4.5	299	36	8.3	1.8
219	4	273	24	11.4	2.8
219	4	248	20	12.4	3.1
219	4.5	416	30	13.9	3.1
219	4.5	357	28	12.7	2.8
219	4.5	339	26	13.0	2.9
219	5	405	32	12.7	2.5
219	5	396	30	13.2	2.6
219	5	377	26	14.5	2.9
219	5	356	24	14.8	3.0
273	5	346	28	12.4	2.5
273	5	330	26	12.7	2.5
273	5	306	24	12.7	2.5
323	5.6	464	32	14.5	2.6
323	5.6	432	28	15.2	2.7
323	5.6	387	24	16.0	2.9

). During the tests, the average frequency was 111 kHz ± 7 kHz depending on the compensation setting, i.e., the ratio of voltage U to current I. The developed mathematical models were also implemented in a spreadsheet in which power P [kW] can be calculated based on the wall thickness and the assumed process speed as well as the selected impeder resulting from the tube diameter.

Table 1 shows the actual data: the wall thickness t and tube diameter, production speed V [m/min] and power P [kW] and linear energy presented as power factors P/V and P/(V·t) used during welding in Huta Łabędy, thereby characterizing the process from the energy side for which high-quality welds confirmed by ultrasonic tests are achieved (98% of production without internal defects). Based on this data, a model was developed to calculate the production capacity P [kW] sufficient to obtain a high-quality weld without internal discontinuities.

2.3. Test Material

The test material consisted of samples taken from welded tubes with production parameters corresponding to the calculation model.

Non-alloy carbon steel tubes for pressure applications, grade P235GH, from a single coil supplier were used for the study. The chemical composition is shown in

Table 2. P235GH steel tube ladle analysis [%w].

Element %	C	Mn	P	S	Si	Ce
Content	0.13	0.45	0.012	0.004	0.006	0.206

and the mechanical properties of the feedstock and tube are shown in

Table 3. Mechanical properties of the feedstock (coil) and pipe P235GH.

Sample Source	Re [MPa]	Rm [MPa]	A5 [%]
Coil	269	408	36.0
Tube	314	423	34.1
Coil	269	417	34.0
Tube	279	412	36.9

.

Both the chemical composition and mechanical properties of the steel met the requirements of the EN 10217-2 standard. High purity steel with 0.004% sulfur content, 0.2% carbon equivalent and very good weldability was used for the tests.

2.4. Metallographic Test

In order to confirm the effectiveness of the applied HFI welding parameters in accordance with the proposed calculation model, detailed metallographic tests were carried out on a sample tube sized 323.9 × 5.6 mm. Samples were taken from the test tubes and then scanned to determine the zonal structure of the weld, the metal flow line and the microstructure in the individual zones as well as to determine the relationship between the geometrical parameters of the weld. The places of measurement of these parameters are shown in Figure 6.

The subjects of the analysis were the following parameters allowing the determination of the correlation between the dimensions of the weld depending on the wall thickness t and the linear energy of welding:

• Central bonding zone (bond line) widths: f_n, f_o and f_i [mm]
• Heat affected zone (HAZ) widths: h_n, h_o and h_i [mm]

Two other factors important for ensuring the required weld quality were the right shape of the metal flow line and the right angle of bending these lines depending on the squeezing force which extruded impurities from the weld zone to provide the required cohesion of the material. The shape of the metal flow lines resulting from the pressure is shown in Figure 7.

The metallographic tests were also supposed to confirm metallurgical purity of the weld which means absence of MnSiO₃ or Mn₂SiO₄ (oxides created by manganese and silicon diffusion) in the direction of the hot edges of the welded strip (reacting with atmospheric oxygen) [19], absence of FeO (formed in the case of low heat supply) and absence of Fe₃O₄ (occurring when the heat supply is too high) [20].

2.5. Automatic Ultrasonic Testing

The ultrasonic testing was carried out using a basic system (Karl Deutsch, Wuppertal, Germany) consisting of 2 pairs of external and internal heads to detect longitudinal discontinuities in the weld zone (up to approx. 6 mm on each side of the weld) for the entire wall thickness, as shown in Figure 8.

The test area is divided into 2 zones: the test area of the pair of heads set on the reference defect placed on the outer surface and the area of the pair of heads set on the reference defect placed on the inner surface. Both N-type grooves are in class U3 for standard requirements for tube P235GH.

3. Results

3.1. The HFI Welding Calculation Model

Figure 9 shows the dependence of the power factor P/V as a function of wall thickness t. For a better illustration, the graph also shows the diameters of the tubes in the form of blue circles corresponding to their diameters on the scale.

By analyzing these dependencies, it can be noticed that the value of the P/V ratio increases with the increase in wall thickness and diameter. The decisive parameter affecting the P/V ratio is the wall thickness. The diameter of the tube is less important.

The results presented in Figure 9 lie in two distinct groups. This is due to the use of various impeders selected for the different tube diameters. In both cases, the change in the P/V ratio (red lines) is similar and amounts to about 2 [kW∙min/m] with an increase in wall thickness by 1 mm. However, the absolute difference in the value of the P/V ratio when using fittings for smaller and larger tube diameters is 4 [kW∙min/m].

Also, when using a circular impeder 1, the value of the P/(V·t) ratio is practically constant and amounts to 2.05 ± 0.10. On the other hand, when using a horseshoe impeder 2 the P/(V·t) ratio is larger (2.77) but also varies in a small range of ±0.21.

Bearing in mind the above, a universal mathematical model was developed and used to describe individual P = f(t,V) functions both when using impeder set 1 and impeder set 2.

The baseline form of the mathematical model entails a function describing the paraboloid plotted by the general Equation (1) where:

$$\mathrm{P}={\mathrm{a}}_1·\left({\left({\mathrm{a}}_2\mathrm{t}+{\mathrm{a}}_3\right)}^2+{\left({\mathrm{a}}_4\mathrm{V}+{\mathrm{a}}_5\right)}^2\right)$$

(1)

where:

t—wall thickness

V—process speed

a₁, a₂, a₃, a₄, a₅—equation coefficients.

This model provided good results in the initial stage of the study (collected experimental data, Table 1) related to the search for the optimal approximating function, during which the universal computational model of welding power P as a function of wall thickness t and process speed V for both types of impeders was developed (Figure 10).

The good approximation properties of the P = f(V,t) function used for the description allowed us to achieve a mean-square error of 11% for impeder 1 and 14% for impeder 2. Unfortunately, for some data, individual approximation errors were well above 20% which is why we decided to develop separate models for the individual types of impeders used for different tube diameters.

To determine the parameters of Equation (1), the [21] tool was used, allowing for the selection of criteria and the control of the process of matching the function. The numerical values of coefficients a₁–a₅ for impeders 1 and 2 matched using the equation are summarized in

Table 4. Coefficients of Equation (1) for different impeders.

Impeder Type	a1	a2	a3	a4	a5
1	1.185	−2.056	−5.426	0.324	−3.949
2	0.532	−6.462	23.327	0.436	12.820

.

Figure 11 and Figure 12 graphically show the form of the approximating functions against experimental data for the diameter range up to 168.3 mm (impeder 1) and for diameters above 219.1 mm (impeder 2), respectively. By using separate approximation models, the mean-square error was reduced to 8% and 9%, respectively.

In parallel, nomograms were determined for individual tube diameter ranges (Figure 13 and Figure 14). The nomograms were used for practical determination of the optimal power P of the induction welding process for a given wall thickness t at a given speed V.

Then, using the developed mathematical model, the power P necessary for the production of an HFI tube sized 323.9 × 5.6 mm was calculated for the process speed increased to 34 m/min (vs. max. 24 m/min used in practice). For the assumed increased speed of 34 m/min, the power estimated by the model for impeder 2 was 497 kW.

Thus, the process parameters provided by the line manufacturer were optimized in order to obtain repeatability of high quality results for all tube dimensions. The use of this model in practice makes it possible to control the geometry of the weld and improve production yields.

3.2. Metallographic Test Results

The typical microstructure of the weld with geometry measurements is shown in Figure 15.

Table 5. Results of HAZ measurements and their dependence on wall thickness t.

Parameter	ho [mm]	hi [mm]	hn [mm]	fo [mm]	fi [mm]	fn [mm]	α * [°]
Value	3.9	4.1	3.3	0.28	0.28	0.19	51
Value vs. t	~0.7 t	~0.7 t	~0.6 t	~0.05 t	~0.05 t	~0.03 t	-

* α—angle of deflection of the metal flow line at ¼ of wall depth t (as shown in Figure 7).

shows the geometric parameters of the weld.

The further course of the research identified microstructures present in the weld. Figure 16 shows a cross-section of the entire weld with five identified zones differing in terms of the microstructure.

The microstructure of the bond zone, namely the center of weld zone 5, consisted of very fine ferrite grains and fine pearlitic colonies shown in Figure 17. There were no acicular structures here. The ferritic structure in the bond line is on the one hand attributable to a decrease in the carbon content (due, among other things, to the decarburization of the hot edges of the strip in the welding process before they came into contact). On the other hand, the concentration of carbon in the liquid phase is higher than in the solid phase. As a result of squeezing/swelling, the high-carbon liquid phase is extruded in the welding/joining zone while the low-carbon solid phase stays in the bond line. Due to the partially ferritic microstructure, the bond line is also called the “ferritic line” (in the case of carbon steels) [19].

In zone 4 with the dominant influence of mechanical deformation (Figure 17), occurring next to the bonding zone, ferrite grains grew to the standard series 5–6 due to the influence of heat, moving further into the thermomechanical deformation zone 3 (Figure 18) where the influence of heat and the time of its interaction was too small to initiate the grain growth processes. The grains in this zone are very fine 11 (grain size was determined according to the PN-EN ISO 643 standard using the comparative method). The deformation of the metal flow line is visible in these zones. This deviation of the flow line is most pronounced at the face and root of the weld where metal with impurities is squeezed out.

Figure 19 shows the zone of dynamically recrystallized ferrite and pearlite grains while Figure 20 illustrates the native material with the typical ferritic–pearlitic structure with equiaxial grains of 9–10 according to the standard series, without non-metallic inclusions.

3.3. UT Ultrasonic Test Result

The test results are shown below in Figure 21. The graph shows the result of the automatic UT test of a 16 m pipe. The scale is shown at the top of the graph with the pipe number. Between the 7th and 8th meters, there is a visible indication which is acceptable in test class U3. Using the calculated optimal parameters, the welded tube met the weld consistency criterion which was confirmed by the ultrasonic tests so no defects in the form of discontinuity of the longitudinal weld were found in the tested tubes.

4. Summary

The presented model for the calculation of HFI welding power P makes it possible to design such welding parameters for a given diameter and wall thickness that guarantee the achievement of a high-quality weld. The model is effective assuming the constants of the equipment used, i.e., the impeder and inductor and assuming one feed supplier. The model can be used to develop, in subsequent studies, an HFI welding simulator that will take into account additional parameters including the quality of the coil, especially waviness, or the state of preparation of the edges before the welding point. Such a simulator can be used in industry to automatically control the HFI welding process after entering input data by the supervisor: namely the tube dimensions and feedstock parameters. The HFI welding simulator can also be equipped with a module for adjusting the shape and dimensions of the hourglass shape of the weld, i.e., the heat-affected zone, but this would require the implementation of additional tube production parameters, primarily the pressure force of the edges before and at the welding point.

The microstructural analyses focused on the measurements of the shape of the weld, showing its regular and symmetrical form along the horizontal and vertical axes as well as on the selection of the appropriate angle of the metal flow line during the extrusion of impurities. The dimensions of the HAZ hourglass should be standardized, as shown in Table 5 and in Figure 15, depending mainly on the wall thickness and the P/V linear energy of welding which was 17.4 in our test. The width of the HAZ hourglass should be approximately 0.7 × t and the width of the binding line should be 0.05 × t. The metal flow lines should deflect at about ¼ of the wall thickness t, with a deflection angle of at least 50°. The flow lines should be arranged symmetrically on both sides of the midline of the bond which proves the right squeezing force to guarantee the formation of the diffusion bond.

Obtaining this shape of the weld and HAZ and the characteristic metal flow line form, while observing the constraints on linear energy, the HFI will guarantee the highest purity and internal consistency of the weld.

The effectiveness of the computational model is also confirmed by the positive results of the ultrasonic tests showing the weld cross-section free from discontinuities.

5. Conclusions

The developed calculation model of linear energy for welding HFI pipes was tested in production conditions and verified by testing the microstructure of the joint. Based on the tests and studies, we can draw the following conclusions:

(1)

The developed mathematical model allows us to calculate the power P [kW] necessary to weld the pipe in the range of walls from 3.6 mm to 12.7 mm. The model takes into account the production speed in the range of 20 m/min to 40 m/min and 2 diameter ranges, i.e., from 114.3 mm to 168.3 mm and from 219.1 mm to 323.9 mm;

(2)

The power P [kW] calculated by the model made it possible to obtain a link without internal defects, confirmed by automatic ultrasonic tests;

(3)

The microstructure of the joint has a zonal structure: zone 1—central line (composed of a bond line and a matrix with a mostly ferritic structure), zone 2—mechanical deformation (with expanded grain to size 5–6 according to PN-EN ISO 643), zone 3—thermomechanical (with fine grain size 11), zone 4—dynamic recrystallization (with irregular ferrite and pearlite grain) and zone 5—native material (with ferritic–pearlite structure typical for the rolling process with equiaxed grains of size 9–10);

(4)

The symmetrical structure of the joint zone, the size of the HAZ and the proper angle of refraction of the metal flow line are in correlation with the linear energy developed as part of the mathematical model;

(5)

Preliminary tests of the linear energy calculation model confirm the applicability in industrial production and are the basis for further work on the welding simulator, which should also take into account other parameters affecting the quality of the joint.