Aspects of Precise Heat Input Control for High Frequency Welding

Aspects of Precise Heat Input Control for High Frequency Welding Lesley D. Frame Thermatool Corp., East Haven, CT, USA University of Bridgeport, Bridgeport, CT, USA lframe@bridgeport.edu, lframe@thermatool.com Kevin Davis, Olexandra Tupalo, Tom Ignatowski, Mick Nallen Thermatool Corp., East Haven, CT, USA Abstract High frequency welding is a thermo-mechanical process that relies on precise heat input as well as mechanical control as strip edges are heated and forged together to result in a seam weld. Heat input can be defined as a way of characterizing the temperature distribution at the strip edges prior to forging them together. Heat input is affected by several process variables ranging from raw material properties to welder settings and weld area setup. These are summarized in this paper, with special attention on the effects of welder frequency, welder power, line speed, and steel alloy composition on heat input and the resulting weld quality. Frequencies in the range of 100 – 800 kHz are considered. Data from tube mills (including general data and controlled on-the-mill experiments) and laboratory evaluations are included in this paper. Introduction High Frequency Welding The High Frequency (HF) welding process is straight-forward: form a strip into a tubular shape, heat the opposing strip edges with a high frequency alternating current (typically 150 - 400 kHz for steel), and forge the heated edges together by passing the formed strip through a set of weld rolls (aka, the weld box). For many plain carbon steel alloys, the process is forgiving and as straightforward as it sounds. However, when new alloys are introduced, new equipment is installed, or things go wrong, it becomes clear just how complicated the process can be. Both the process parameters involved with HF welding [1-3] and the equations defining precise heat input [49] have been discussed at length and will not be covered here in detail, but it is necessary to revisit a few key aspects of the HF weld process in order to appreciate what is meant by “precise heat input control” and why it is important. The overall quality of the final product is influenced by each of the following four aspects of the HF welding process: Raw Material Selection; Forming; Welding; Post-Weld Processing. However, the weld integrity can be discussed as a function of what goes in (Raw Material Strip) and how the edges of the strip are joined (Welding). This paper summarizes many of the potential challenges for High Frequency welding as well as relevant process parameters that can and should be controlled. Fortunately, advances in steel making are leading to improved microstructural and compositional consistency in raw materials, and new developments in welder technology have enabled enhanced control of the HF weld process. These improvements provide the tube and pipe industry with new confidence in the robust and cost effective method for production that is HF Welding. What follows is a discussion of the fundamental principles of the process. Influence of Raw Material It is not news that the raw material significantly impacts the final quality of the weld and the product as a whole, but the degree to which the raw material plays a role is sometimes mystifying. The primary materials properties of interest for HF Welding are the thermal conductivity (k) and the electrical resistivity (). Figure 1 illustrates the spread in these properties for several alloy systems. For HF welding, the electrical resistivity influences how easy it is to create heat in an alloy through the Joule Heating Effect, and the thermal conductivity of the alloy indicates how quickly that heat will transfer away from the strip edges. For the purposes of this research, the focus is plain carbon and alloy steel. Although Fig. 1 shows single values for k and  for any given alloy, these materials properties are not constant. Both will change as a function of temperature and microstructure. For example, the electrical resistivity of an alloy is inversely proportional to grain size [12,13], and proportional to carbon content. This is important for HF welding because it is possible to have slightly different grain size on the two opposite strip edges or microsegregation of elements like carbon from edge to edge or even surface to core as shown in Fig. 2. Thermal Conductivity and Electrical Resistivity of several alloy groups 450 Plain Carbon Steel Alloy Steel 400 Austinitic Stainless Steel C100 - C400 Ferritic Stainless 350 Thermal Conductivity (W/m-K) Martensitic Stainless C100 Copper alloys 300 C200 Copper alloys C300 Copper alloys 250 C400 Copper alloys C500 Copper alloys 200 C600 Copper alloys Al alloys C700 Copper alloys 150 C800 Copper alloys Plain and alloy steel C900 Copper Alloys 100 C500 - C900 1000 Series Al Stainless Steel 2000 Series Al 50 3000 Series Al 4000 Series Al 0 5000 Series Al 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 6000 Series Al Resistivity (W-m) Figure 1. Thermal Conductivity and Electrical Resistivity of several alloy systems (compiled from [10,11]). Welding Parameters The material properties and microstructure of the raw material will influence how the strip heats up once current is introduced, but there are thermal, physical, and electrical aspects to how the heat is introduced and controlled. HF welding relies on electromagnetic phenomena (Skin Effect, Proximity Effect, and Joule Heating Effect), heat transfer principles (thermal conduction), and metallurgical phenomena (phase transformation, dynamic recrystallization, dynamic grain growth). Table 1 provides simplified mathematical relationships or descriptions for each of these phenomena and lists how each can be controlled during HF Welding. For a more thorough discussion of how these phenomena relate to heat input in High Frequency welding, refer to the extensive work by Dr. Paul Scott [4-7,9]. The key to understanding and controlling heat input stems from first understanding the interaction of these process parameters. As shown in Table 1, aspects like line speed, v, and vee geometry (Lvee and vee) feed into several different fundamental relationships and phenomena. By examining the influence of individual parameters, it is possible to rank the relative impact of these parameters and better understand heat input for HF Welding. In the study reported here, the focus was on power (current), frequency, line speed, vee length, vee angle, alloy composition, and wall thickness. Table 1. HF Weld parameters Description of Phenomena Joule Heating Effect 𝜌(𝑇) ∙ 𝐿(𝐿𝑣𝑒𝑒 , 𝜃𝑣𝑒𝑒 ) 𝑃 = 𝐼2 ( ) 𝐴(𝑥𝑤𝑎𝑙𝑙 , 𝜉) 𝐶𝑝 (𝑇) ∙ 𝑚(𝐿, 𝐴) ∙ (𝑇𝑓 − 𝑇𝑖 ) = 𝑡(𝐿𝑣𝑒𝑒 , 𝑣) Skin Effect 𝜌(𝑇) 𝜉=√ 𝜋 ∙ 𝑓 ∙ 𝜇𝑟 (𝑇, 𝐵) Proximity Effect Opposite current in adjacent conductors will concentrate at the surface closest to the adjacent conductor. Thermal Conduction 𝑇 − 𝑇𝑠 𝑇𝑖 − 𝑇𝑠 𝑥 = erf ( ) 2 ∙ √𝛼(𝑘, 𝜌, 𝐶𝑝 ) ∙ 𝑡(𝐿𝑣𝑒𝑒 , 𝑣) Phase transformation, Dynamic recrystallization and grain growth (T, Ṫ, v, F, ) HF Weld Parameters Raw Material: : electrical resistivity Cp: Heat Capacity xwall: wall thickness Controllable Settings: P: Welder power Lvee: Vee Length vee: Vee angle v: Line speed Raw Material: : electrical resistivity r: relative magnetic permeability Controllable Settings: f: frequency Raw Material: : electrical resistivity Controllable Settings: f: frequency vee: Vee angle P: Welder power Raw Material: k: thermal conductivity : density Cp: Heat Capacity Controllable Settings: Lvee: Vee Length v: Line speed Raw Material: Prior microstructure Controllable Settings: v: line speed F: Forge pressure Methods Weld Line Data Collection This study considers high frequency welding process data from 92 different HF welding mills around the world and 280 distinct production runs as well as the data and corresponding metallurgical weld samples from one particular mill (33 samples). In this study these are termed “unconstrained” and “constrained” data, respectively. The unconstrained data were collected by Thermatool service engineers and field technicians on a wide range of mills with varied welder models and for different alloy compositions. One commonality with these data is that they represent each mill’s optimized settings (based on operator and mill plant judgement). As discussed below, this makes this data set useful only as an aggregate and with reservation. Fine Grain High resistivity Coarse Grain Low resistivity (a) The constrained data and 33 samples were collected on a mill using an HF welder with controllable frequency (a Thermatool HAZControlTM Welder), with carefully adjusted parameters. Three steel alloy types were run during these trials (AISI 1010, 1020, and 4140 steel). During each trial, Thermatool engineers, mill engineers, and mill operators recorded line speed, power setting, frequency, vee angle, and vee length, along with the product dimensions and composition. Forge pressure was not measured during the trials, but the amount of upset (distance between weld rolls) was held constant, allowing for the assumption that displacement during forging is a fixed parameter. These experiments were useful as a means to compare the controllable HF weld parameters despite the fact that the resulting weld quality for most samples was quite poor. OD Comparing Data to Simulation Results In addition to the analysis of the mill data and HF weld samples, simulation using COMSOL Multiphysics provided opportunities for comparison to and interpretation of the results. Mid Metallurgical Analysis HF weld samples from the constrained dataset were hot mounted and polished to 1m before etching with 2% nital for 20 seconds to reveal microstructure. Images were processed and analyzed using ImageJ software and Adobe Photoshop. ID 0 25 Data Collection 50m (b) Figure 2. Examples of asymmetry in steel strip. (a) Differences in grain size and carbon microsegregation between left and right of steel strip, and (b) differences in carbon microsegregation from OD to ID of strip. Unconstrained Data Because the unconstrained data was collected under a wide range of conditions, by at least a dozen different people, using different tools, and for different alloys, it is not possible to compare the mass of data without several considerations and caveats. One important finding is that the relationships between different process variables can easily lead to false conclusions. For example, Fig. 3 illustrates the unconstrained steel data when only two process variables are compared. Clearly bidirectional relationships comparing power, wall Power (kW) Power (kW) Power vs Wall Thickness (Steel) 1000 900 800 700 600 500 400 300 200 100 0 0.0 0.2 0.3 Wall Thickness (in) 0.4 0.5 0.0 Frequency vs Wall Thickness (Steel) 400 350 350 300 250 200 150 100 100.0 200.0 300.0 400.0 Line Speed (fpm) 500.0 600.0 Frequency vs Line Speed (Steel) 450 400 Frequency (kHz) Frequency (kHz) 450 0.1 Power vs Line Speed (Steel) 1000 900 800 700 600 500 400 300 200 100 0 300 250 200 150 100 50 50 0 0 0.0 0.1 0.2 0.3 Wall Thickness (in) 0.4 0.5 0.0 100.0 200.0 300.0 400.0 Line Speed (fpm) 500.0 600.0 Figure 3. Two-dimensional representation of the HF weld process parameters is not sufficient to illustrate the relationships. thickness, line speed, and frequency are not sufficient to tell the whole story. It is necessary to consider relationships in three or more dimensions as illustrated in Fig. 4 for the patterns to really emerge. It is problematic and misleading to consider only two variables at a time when attempting to understand heat input for HF welding. Although simplifications like this have been attempted in the past by examining only frequency and vee angle [14-16], one should be cautious of studies that overlook the multidimensional aspects of HF weld process parameters. Furthermore, Fig. 4 and the results of the research presented below suggest that there is significant value in being able to monitor and control each of these process parameters independently of each other in order to maximize control over heat input to the strip edges during HF welding. Relationship between line speed, power setting, frequency selecting, and wall thickness for steel products Frequency (kHz) 160 170 181 191 202 212 223 233 244 254 265 275 286 296 307 317 328 338 349 360 370 381 391 402 412 423 Speed (fpm) ​ ​ ​ 700 600 ​ 500 ​ 400 ​ 300 ​ 200 ​ 100 ​ ​ ​ ​ ​ ​ ​ 0.09 ​ 0.18 Wall (in) ​ 0.27 ​ ​ 0.36 200 400 600 Power (kW) 800 ​ 0.45 Figure 4. Four-dimensional representation of all unconstrained HF welding data with frequency shown by marker color. The three axis chart in Fig. 4 illustrates the tendency to use lower power at faster line speeds and thinner walls, but as wall thickness increases there is much greater spread in the data. Frequency is shown by the bubble color, and it should be noted that these data reflect the parameters that were selected by the mill operators to achieve a sellable product. The steel composition covers a wide range from HSLA to alloy steel, and the applications include mechanical tubing, structural, API, refrigeration tubing, automotive, furniture, and others. For these reasons, these data can only provide a big picture view of the range of variability in the process parameters. Furthermore, frequency in these data is not necessarily a good indicator of the optimal frequency for the particular products because many of these mills did not have a selectable frequency welder (these were primarily fixed frequency welders). It is likely that the pattern observed in Fig. 4 showing higher frequency for the thinner wall applications is more representative of plant purchasing decisions than mill operator frequency selection. That is, there were other influencing factors that drove the plant to purchase a 200 kHz or 400 kHz welder, and by the time the operator was able to select welding parameter settings, frequency had already been effectively selected by others. When only the plants with variable frequency welders are displayed (n=50), the pattern for frequency is no longer present (Fig. 5). From these data it is impossible to determine the reason for the lack of pattern. Constrained Data The constrained welder data is much more limited in number (n=33), but in many ways it shows a richer picture of the relationships because many more parameters were recorded and controlled. Figure 6 shows this dataset overlaying the unconstrained steel data. Despite the fact that the welding parameters selected for the constrained dataset do not necessarily reflect the optimized conditions (many of the resulting welds would not pass quality inspections), the data points plot amidst the unconstrained dataset, making this a reasonable sample set to compare to the larger picture. Relationship between line speed, power setting, frequency selecting, and wall thickness for steel products on variable frequency welders amount of strip edge material that is squeezed out as the tube passes through the weld forge rolls. Frequency (kHz) Speed (fpm) 400 ​ 320 ​ 240 ​ 160 ​ ​ ​ ​ ​ 140 210 ​ ​ 0.14 Wall (in) ​ 0.21 ​ 0.28 70 ​ ​ ​ 0.07 Power (kW) 280 ​ Flow lines 80 ​ ​ Parent Material 208 216 223 231 239 247 255 262 270 278 286 293 301 309 317 324 332 340 348 355 363 371 379 386 394 402 Centerline (ferrite line) 350 ​ 0.35 Squeezed out Coarse-grained HAZ Figure 5. Four-dimensional representation of unconstrained HF welding data from only the mills using variable frequency welders. Line speed Unconstrained ​ ​ ​​ ​ 700 ​ 600 ​ 500 ​ 400 ​ 300 ​ 200 ​ 100 ​ ​ ​ ​ 0.09 ​ 0.18 ​ 0.27 Wall (in) ​ ​ 0.36 200 400 600 2mm Two-phase HAZ Figure 7. Schematic drawing of the features of a typical steel HF weld and a corresponding micrograph of an HF weld. Constrained and Unconstrained Datasets Constrained 0 Fine-grained HAZ Power (kW) 800 ​ 0.45 The HAZ for HF welds experiences a thermal gradient with the highest temperatures at the strip edge and corners. In Fig. 8, the strip edge thermal gradient is compared to an equilibrium Fe-C phase diagram to illustrate the phases present in each region of the HAZ. The temperature indications shown in Fig. 8 are examples of what can typically be achieved during welding, not precise temperature targets. Also it should be noted that the high frequency welding process is by no means an equilibrium process, and the comparison to an equilibrium phase diagram is meant to provide a loose point of comparison. The exact temperatures for Ac1, Ac3, and grain coarsening depend on both composition and heat rates. However, the illustration in Fig. 8 does make it relatively easy to see that the portion of the strip edge that is squeezed out during forge welding is mostly made up of coarse grain austenite. Therefore the presence and size of the coarse grain zone within the HAZ is especially dependent on Figure 6. Constrained data shown in relation to unconstrained dataset. Metallurgical Analysis High Frequency welds in steel alloys often follow the form shown in Fig. 7. Although not all features need be present in all HF welds, and the absence of specific features (e.g., the “ferrite line” or “flow lines”) does not indicate poor weld quality. The examination of HF weld involves characterization of flow lines, heat affected zone (HAZ) width, metallurgical phases, grain size, and HAZ symmetry. The HAZ can include a coarse grain region near the weld line, a finer grain region adjacent to that, and a two-phase region at the edges of the HAZ. The presence and width of these regions within the HAZ are influenced by the steel composition, the temperatures reached during heating, and the 3 2 1. 2. 3. Two-phase region Fine-grain region Coarse-grain region 1 1 2 3 HAZ expanded for clarity. Figure 8. The relationship between the strip edge thermal profile and the resulting phases. the weld box settings (how much material is squeezed out). If the amount of upset (squeeze out) is not recorded, it is not appropriate to compare HAZ width results when degree of upset is changed. In this study, upset was not recorded, but it was held constant for all weld samples produced. The HAZ’s for some of the samples analyzed are shown in Fig. 9. The steel composition and wall thickness clearly play large roles in the nature of the HAZ. Upon closer examination, however, it is possible to discern the distinct regions of the HAZ for both the plain carbon steel examples (Fig. 10) and the alloy steel examples (Fig. 11). HAZ Width The coarse grain (CG) HAZ width and full HAZ width were measured for all of the samples in this study. In Fig. 12, the coarse grain HAZ width is plotted against frequency. The electrical reference depth as a function of frequency is also shown (note that the electrical reference depth has been 1010, Sample #4656 1020, Sample #4654 4130, Sample #4629 4130, Sample #4649 0 200m Figure 9. HAZ macrographs for 1010, 1020, and 4130 steel samples from the present study (scale is the same for all images). HAZ Parent Fine Grain Two-Phase Coarse Grain 1020, Sample #4621 0 200m Figure 10. HAZ micrograph for 1020 steel showing the parent microstructure, two-phase region, fine-grained HAZ, and coarse grain HAZ. Note, this represents a weld without optimized HF weld process parameters. HAZ Parent Two-Phase Fine Grain Coarse Grain 4130, Sample #4627 0 50 100m Figure 11. HAZ micrograph for4130 steel with parent microstructure, two-phase region, fine-grained HAZ, and coarse grain HAZ. Coarse Grain HAZ width - 1020 30 40 50 60 70 80 90 100 110 120 130 400 350 300 250 200 0.5 1 1.5 2 Coarse Grain HAZ width (mm) 2 x Electrical Reference Depth (mm) 1020-Frequency Electrical Reference Depth (Calc) Power (kW) Frequency (kHz) 450 2.5 0 200m 1020 - Power Frequency (kHz) 30 50 70 90 110 130 150 170 190 400 350 300 250 200 0.5 1 1.5 2 Coarse Grain HAZ width (mm) 2 x Electrical Reference Depth (mm) 1010 - Frequency Electrical Reference Depth (Calc) Power (kW) Coarse Grain HAZ width - 1010 450 4130, Sample #4648 (top) and #4649 (bot) 2.5 1010 - Power 30 400 40 50 350 60 300 70 250 Power (kW) Frequency (kHz) Coarse Grain HAZ width - 4130 450 80 200 90 0 0.5 1 1.5 2 Coarse Grain HAZ width (mm) 2 x Electrical Reference Depth (mm) 4130 - Frequency Electrical Reference Depth (Calc) 2.5 4130 - Power Figure 12. Charts showing the coarse grain HAZ width for all samples against frequency and power. doubled for these charts because we are looking at both strip edges forged together, not only one half of the weld). The coarse grain region falls well below the electrical reference depth for all samples. This suggests that the Joule Heating effect is responsible for directly heating everything that was squeezed out as well as at least some (if not all) of the coarse grain region. The rest of the HAZ was likely not heated directly by the Joule Heating Effect, but without a measurement on the amount of upset, it is not possible to determine exactly how much of the non-Coarse Grain HAZ was heated directly by the Joule Heating Effect. Regardless, the charts in Fig. 12 illustrate an inverse relationship between CG-HAZ width and frequency and a weak (if any) relationship with power. The conclusion that the coarse grain HAZ is primarily due to the Joule Heating Effect is further supported by Fig. 13. COMSOL modeling illustrates the current density at the strip edge at 200 kHz (bottom) and 400 kHz (top) for 4130 steel with 0.1” wall thickness. Corresponding macros of the 4130 samples (0.13” wall) are shown. The depth of current flow on the strip edges is greater than the coarse grain region shown in the macros. Figure 13. COMSOL model of current density at the strip edge of a 4130 steel tube with 0.1” wall at 400 kHz (top) and 200 kHz (bottom) with corresponding weld samples. Accepting that the CG - HAZ is primarily due to the Joule Heating effect, and the full HAZ is wider than the electrical reference depth at the strip edges could account for, then the fine grain and two-phase regions of the HAZ must be due to a combination of the Joule Heating effect and thermal conduction of heat from the strip edge. As indicated in Table 1 above, thermal conduction is a function of time and several materials properties, which change with temperature and are not controllable with the process variables. Time is controllable in this situation. Figure 14 (top) shows the relationship between the HAZ width without the coarse grain region (considering only the fine grain and two phase regions) versus vee time. Vee time is a parameter that indicates the amount of time that any strip element spends in the vee as it heats up. It is equal to the vee length divided by the line speed. This is a way to take into account changes in vee length and line speed among all samples. The 1010 steel samples were only run with a limited set of processing conditions, so no relationship is apparent in that sample set. However, both the 1020 and 4130 samples illustrate a relationship between the non-coarse grain HAZ width and vee time. Longer time in the vee will result in a wider two phase and fine grain HAZ region. The coarse grain HAZ does not illustrate the same relationship (Fig. 14, bot). This supports the hypothesis that these regions are primarily influenced by thermal conduction (so-called, “thermal mode” heating) and the coarse grain HAZ is primarily influenced by the Joule Heating Effect (so-called, “electric mode” heating) [4]. Microstructure Microstructure of the coarse grain HAZ for the 1010, 1020, and 4130 steel samples are shown in Figures 15 – 17, respectively. During HF welding, the HAZ cools quickly after it passes through the weld forge rolls due to the presence of mill coolant and also the heat sink effect of the rest of the tube. Therefore, the microstructure in the HAZ is typically martensitic when there is sufficient carbon in the steel. For the 1010 steel, the microstructure is mostly ferritic. 1020 Two-Phase HAZ width (mm) Effect of Line Speed and Vee Length on HAZ Width 2.2 2 1.8 1.6 1020 1.4 4130 1.2 1010 1 0.8 0.1 0.15 0.2 0.25 0.3 Vee Time (sec) 0.35 0.4 1020, Sample #4624 25 50m Figure 16. Coarse grain HAZ region for sample #4624, 1020 steel. Effect of Line Speed and Vee Length on CG HAZ Width CG HAZ width (mm) 0 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0 1020 4130 1010 0.1 0.15 0.2 0.25 0.3 Vee Time (sec) 0.35 0.4 Figure 14. Vee time (a function of line speed and vee length) versus HAZ width. 4130, Sample #4647 0 25 50m Figure 17. Coarse grain HAZ region for sample #4647, 4130 steel. 1010, Sample #4635 0 25 50m Figure 15. Coarse grain HAZ region for sample #4635, 1010 steel. Prior austenite grainsize was measured in the CG HAZ region for 1020 and 4130. The relationship between grain size and frequency in Fig. 18 shows larger prior austenite grain size with higher frequency. Although the power was not held constant for all of these weld runs, when the power is normalized against changes in line speed and wall thickness, the power settings for all weld samples shown in Fig. 18 are within 4% of each other. From a logical consideration of the phenomena at play, the relationship displayed in Fig. 18 is not surprising. With higher frequency, the current density at the strip edge would be higher, allowing the strip edge material to reach a higher temperature. Grain coarsening takes place very rapidly above the grain coarsening temperature [17], so it does not take a very large temperature change to see a very large change in grain size. Grain Size vs Frequency Frequwency (kHz) 450 400 350 1020 300 4130 250 200 25 30 35 40 45 Grain Size (m) Figure 18. Effects of power and frequency on prior Austenite grain size. Conclusions This study draws from a large set of field data, but the more useful dataset is the smaller collection of samples from a series of controlled on-the-mill experiments. The examination of heat affected zone (HAZ) width and prior austenite grain size in the coarse grain HAZ region of the welded samples supports the following conclusions: 1. The coarse grain heat affected zone is primarily the result of direct heating by the Joule Heating Effect and there is a strong relationship between the welder frequency and coarse grain HAZ width and a weaker relationship between welder power setting and coarse grain HAZ width. This has been referred to as “electric mode” heating by others [4]. 2. The relationship between frequency and coarse grain HAZ width is also affected by alloy composition and power setting with a stronger relationship between frequency and CG HAZ width at higher power settings (for the same alloy), and narrower CG HAZ for the alloy steel. 3. The fine grain and two-phase HAZ regions are primarily the result of thermal conduction from the strip edge and the width of these HAZ regions is heavily influenced by vee length and line speed regardless of frequency and power. This has been referred to as “thermal mode” heating by others [4]. 4. The prior austenite grain size in the CG HAZ is a function of frequency. Although these results provide insight into the HF welding parameter relationship, more research is necessary to fully quantify and verify the patterns that have emerged. On-themill experimentation is time consuming and costly when one is not concerned with producing sellable product. However, if the following parameters are recorded during normal production runs, it would be possible to compare disparate datasets to further fill in the gaps in our assessment. The key parameters are: power setting, frequency at the work coil (measured), vee length, vee angle, line speed, alloy composition, product dimensions, and amount of upset during forging. There are other parameters and factors including the weld box design, use of impeders, and placement of coolant in the weld area that can also be explored. It is especially important to note that these results are not meant to indicate that high frequency is always “good” or “bad.” Rather, these data show that frequency can be a powerful tool for precise heat input control when concerned with weld microstructure and grain size and used in conjunction with a well-controlled and stable HF weld process. Furthermore, characterizing and understanding the relationship between the main controllable parameters for HF welding will lead to better process control and better overall product quality. Acknowledgments Thermatool Corp. would like to thank their customers who have allowed data collection and on-the-mill experimentation to allow better understanding of the interaction of HF Welding process parameters. References [1] Frame, L. et al., “Factors Affecting Grain Size in High Frequency Welding,” MS&T 16, Salt Lake City, UT, October 24-27, 2016. [2] Frame, L., “Controlling HF Weld Quality Part 2: Characterizing the weld and troubleshooting weld defects,” FABTECH, Las Vegas, NV, November 2012. [3] Frame, L., “HF Welding Process and Weld Quality Metrics,” 3o Semenario Technologia en Fabricación de Tubos, Itu, Brazil, October 2012. [4] Scott, P. and Smith, W., “Key Parameters of High Frequency Welding,” Tube International, Vol. 15 (1996), pp. 147-152. [5] Scott, P., “The Effects of Frequency in High Frequency Welding,” Tube 2000 Toronto, ITA Conference, Toronto, ON (1996). [6] Scott, P., “High Frequency Welding of Low Carbon Steel Tube,” Thermatool Corp., East Haven, CT. [7] Scott, P. “Controlling the Heat Affected Zone (HAZ) in HF Pipe and Tube Welding,” Society of Manufacturing Engineers, (2000). [8] Gardiner, D., “Variable Frequency on Demand – The Ultimate in Flexibility for Today’s Tube and Pipe Producers,” Tube & Pipe Technology Magazine, July/August (2003), pp. 82-83. [9] Scott, P., “Selecting a Welding Frequency: Research on the Optimal Frequencyt for Tube Production,” The Tube and Pipe Journal, October/November (2003). [10] Cverna, F., ASM Ready Reference: Thermal Properties of Metals, ASM International (Materials Park, 2002). [11] Moosbrugger, C., ASM Ready Reference: Electrical and Magnetic Properties of Metals, ASM International (Materials Park, 2000). [12] Arajs, S. et al., “Effect of Grain Boundaries on the Electrical Resistivity of High-Purity Iron at 4.2°K,” Journal of Applied Physics, Vol. 38, No. 4 (1967), pp. 1676-1677. [13] Karolik, A.S. et al., “Calculation of electrical resistivity produced by dislocations and grain boundaries in metals,” Journal of Physics: Condensed Matter, Vol. 6, No. 4 (1994), pp. 873. [14] Grande, B. and Asperheim, J.I., “Factors Influencing Heavy Wall Tube Welding.” Tube & Pipe Technology Magazine, March/April (2003), pp. 86-88. [15] Asperheim, J.I and Grande, B., “Temperature Evaluation of Weld Vee Geometry and Performance,” Tube International (UK), Vol. 19, No. 110 (2000), pp. 497-502. [16] Grande, B., et al., “Weld setup, variable frequency and heat affected zones in high-frequency tube and pipe welding,” Tube & Pipe Technology Magazine, July (2012), pp. 116-119. [17] Al-Hajeri, K., The Grain Coarsening and Subsequent Transformation of Austenite in the HSLA Steel During High Temperature Thermomechanical, PhD Dissertation, University of Pittsburgh (Pittsburgh, 2005).