Article collection: Tube and pipe welding

Article

Weld setup, variable frequency and heat affected zones in high- frequency tube and pipe welding By Bjørnar Grande, Olav Wærstad and Peter Runeborg (EFD Induction)

Introduction Tube and pipe manufacturers aim to achieve and repeat successful production runs – something requiring knowledge of the impact of many parameters in the manufacturing process. One of the first theoretical research works based on Finite Element Method calculations on the two-dimensional (2D) heat affected zone (HAZ) was published in 1998, with the focus on weld frequency [1] . Further studies focused on geometrical variables of the weld vee [2, 3] . A key result of this research was the realisation that geometrical parameters have a significant impact on the HAZ. This suggests that more attention should be paid to weld setup control in order to obtain the desired HAZ. In addition to the use of welder recipes (tube identification, power set point, energy monitoring factor, etc) weld setup recipes should be used to maintain the HAZ for all production batches of a product [9] . Other published research focused solely on the weld frequency’s impact on the HAZ, and has resulted in a proposed welder concept that includes frequency adjustment to control the HAZ [6, 8] . This paper, from a principal point of view and based on a 2D model of the HAZ, investigates the proposed concept’s ability to repeat a product’s HAZ throughout production. The article investigates the impact that geometrical changes in the weld zone have on weld frequency and the Heat Affected Zone (HAZ). The article evaluates the consequences of controlling HAZ by a variable frequency option. The article points out the importance of weld setup control.

a) 2.8mm wall, steel

b) 8.9mm wall, steel

Figure 1: Real 2D heat affected zones

(0.11") and 8.9mm (0.35"), for two common steel materials. The hourglass shape of the HAZ is clearly visible, showing that the heating of the faying strip edges is not uniform across the wall thickness. HAZ control concept One main objective of the proposed HAZ control solution is to reproduce the HAZ of an earlier production run [5, 6] . The proposal makes two separate but related claims: 1) It is possible to calculate the 1D temperature distribution in the x-direction, and the maximum vee wall surface temperature at x=0, provided we know certain tube material properties, the weld speed and the weld vee length (Figures 2a and 2b) 2) Weld frequency and welder output power can control the 1D temperature distribution in the x-direction, and the maximum vee wall surface temperature at x=0 With the ability to estimate the shape of this heat distribution, the HAZ width can be calculated and controlled. The HAZ width is given by the temperature assumed as the lower limit of the HAZ. It is denoted ½ * HAZ in Figure 2b, since the total HAZ is given by the area of heated material on both tube wall edges.

Heat affected zone The heat affected zone is typically defined as the area of base metal where the microstructure and material properties have been altered by the welding process and subsequent re-cooling. One author defines the HAZ as any metal heated to 650°C (1,200°F) or hotter [4] . Figure 1 shows two weld samples, wall thicknesses of 2.8mm

Figure 2a: System of axes

Figure 2b: Temperature distribution

116

www.read-tpt.com

July 2012

Article

The final step evaluates how this system response affects the HAZ, which has already been altered by the initial change in the weld set-up (compared with the previous production run). This lets us evaluate the overall value of the proposed system.

If the calculated HAZ width is not as requested, the weld operator or a computer must adjust frequency according to the following: • Calculated HAZ width < requested HAZ width reduce frequency • Calculated HAZ width > requested HAZ width increase frequency In addition, welder output power is adjusted by the operator or calculated by a computer program to give the required energy input to obtain the requested HAZ width and weld temperature at the vee wall surface (x=0). The HAZ at the weld point is the most significant parameter, so T(x) is calculated at y = vee length. The whole concept is based on uniform current distribution in the weld vee walls and a uniform temperature across the weld; that is, a 1D model [8] . Parameters influencing the HAZ and investigation procedure A number of parameters influence the weld temperature and heat distribution in the weld vee, thereby affecting weld quality. Loebbe presents 16 such parameters [7] . Focusing on the HAZ and the geometrical parameters that can change over time in the weld zone, we examine the following: • Weld vee angle and springback • Moving weld point, continuously changing position • Non-stable vee angle (‘breathing’ vee), continuously changing vee angle • Distance weld point – coil (or contacts); the vee length One idea critical to the proposed HAZ control concept is to reproduce from an earlier production run the temperature distribution and the maximum weld temperature at the tube wall’s surfaces. In the proposed concept, the two parameters to be adjusted (by the operator or computer system) are the welder frequency and welder power. The first step is, therefore, to determine how changes in the weld setup parameters alter the resonance circuit’s frequency and the required load power. This is what we call the process response . The second step is to identify the adjustments of welder frequency and welder power, according to the proposed control concept. This can be called the system response .

Process response The resonance frequency for both series and parallel resonance circuits is given by:

Note: Valid for both current-fed and voltage-fed inverter-based welders

C is the total capacitance of the electrical circuit and is given by the installed compensating capacitors inside the welder’s cabinet. L i is the internal inductance of the welder and consists of the inductance in coil leads, busbar and the output circuit parts inside the machine’s cabinet. L Load is the load inductance and, in the case of induction welding, can be divided in three parts: OD tube ; mainly due to air gap between the induction coil and the outside surface of steel strip • L

• L

; mainly due to air gap between the strip edges in the weld vee ; mainly due to impeder and air gap between impeder and inside surface

Vee

• L

ID tube

The two last inductances are in parallel in the equivalent electrical circuit (Figure 3). The process responses are listed in Table 1. It is important to note that these responses are independent of welder type. These are the process responses. The symbol ‘-’ denotes no change.

Figure 3: Electrical circuit model of tube (strip)

Table 1: Process responses to parameter changes

Parameter

Change

L

L

L

L

Frequency Power

vee

IDtube

ODtube

Load

Wider

Inc(rease) Dec(rease)

- - - - - - - -

- - - - - - - -

Inc

Dec

Inc

Vee angle

Narrower

Dec

Inc

Dec

More Less

Inc

Inc

Dec

Inc

Springback

Dec

Dec

Inc

Dec

Downstream Inc Upstream Dec

Inc

Dec

Inc

Moving weld point

Dec

Inc

Dec

Longer Shorter

Inc

Inc

Dec

Inc

Vee length

Dec

Dec

Inc

Dec

117

www.read-tpt.com

J uly 2012

Article

DNSA mill: Turks head Weld vee angle Figure 4 summarises the results of previous research [1, 2, 3] . A wider vee angle and higher frequency have both the same principal impact on the HAZ. In resonance frequency, the process response to a wider vee angle is a lower frequency (Table 1). The wi er vee angle results in a more pronounced hourglass-shaped HAZ (Figure 4). To keep the calculated HAZ width unchanged, the proposed system’s response to a wider vee angle is to increase the weld frequency to that of the reference production run. But, according to Figure 4, this response does not compensate for the initial change in HAZ, as would be expected for a control system. On the contrary, it amplifies the change by increasing the heating of the corners. The outcome of this situation, with the HAZ control concept, is that production continues to run, at a somewhat higher power (within accepted tolerances), with a more pronounced and amplified hourglass-shaped HAZ than for the reference run. A welder without the step-less variable frequency option also continues production with the changed HAZ shape, with the somewhat higher power output, but at the lower frequency (process response). The initial change in HAZ, due to the wider vee angle, is somewhat counteracted by the natural reduction in frequency (Figure 4). Figure 5 summarises these effects. In case the initial parameter change is a narrower weld vee angle, the process response, the system’s response and change in HAZ are all in the opposite directions. The final outcome is, however, the same as outlined above. System response The investigation of the system response and its influence on the HAZ starts with a change in the weld vee angle. First, the proposed HAZ control concept [5, 6] is investigated, followed by an evaluation of a welder design (with constant internal inductance) without the feature of step-less adjustable frequency. In the proposed system, power input is one of two parameters to be adjusted. If power can not be adjusted within set tolerances of the reference run, the operator must check and adjust weld setup to get the same power as in the previous production run. This is exactly the same procedure as for a plain welder that does not include the proposed concept. All welders should have a feature such as a weld recipe for each product to be welded [9] . Power can then be compared against a recorded value for the reference product. Further investigation of frequency adjustment and HAZ control, therefore, will be based on required power being within set tolerances after a change in the actual parameter.

Figure 5: Implications of change to wider vee angle for the HAZ

Springback The resilient flexing of the strip edges, also called springback, was investigated with respect to both heating of the corners of the strip and depth of heating in the centre of the tube wall [2] . The more pronounced the springback, the more pronounced the hourglass shape of the HAZ. The springback has the same influence on HAZ shape as the vee angle. As can be seen in Table 1, the process response in the springback case equals that of the weld vee angle. The system response of the proposed HAZ control concept will then be identical to the one described above for the vee angle. The final result is that no real HAZ control has been achieved by the proposed HAZ control concept, which amplifies the initial change in the temperature distribution. Moving weld point and breathing vee Situations with a moving weld point and the ‘breathing vee’ are most likely to occur at the same time. As the weld point moves downstream and upstream, the distance between the strip edges increases and decreases accordingly. This is a situation with a continuously changing effective weld vee angle and length. Analysis of the results shows that the heat penetrates deeper into the material in circumferential direction both in the centre of the tube wall and at the outside and inside surfaces as the vee length increases [2] . The shape of the HAZ is relatively unchanged as the vee length (heating time) changes. But the overheating of the corners occurs as the heating time gets longer [3] . First, we assume that the weld point moves downstream. According to Table 1, the process response to an increased vee length is a lower frequency. Due to the request to keep calculated temperature distribution equal to the distribution of the reference run, the proposed system’s response is to increase frequency. The initial, extra heating of the corners is strengthened by the increase in frequency. Next, when the weld point moves upstream, the opposite will happen. This means that the proposed system continuously amplifies the mechanically initialised HAZ changes. Weld vee length The distance from the weld point to the coil is an input value to the HAZ control system. Therefore, the evaluation in this

Figure 4: HAZ shape vs frequency and vee angle

DNSA mill: Guide roll and impeder

118

www.read-tpt.com

July 2012

Article

Summary The above evaluation focuses on the impact of geometrical parameters on theHAZ, and the influence frequency adjustment has in maintaining the HAZ when the weld vee geometry changes. Other parameters, such as coil and impeder, are not covered. The general result of this investigation is that the inherent response of the HAZ control system is to amplify the initial HAZ change caused by geometrical alterations, rather than opposing changes, which would be an expected and desired response of a control system. This 2D-model-based investigation shows that the ability to adjust frequency as described in the proposed concept can not compensate for changes in HAZ shape caused by geometrical changes in the weld zone. For the investigated parameter changes, the inherent property of the system is to ensure a constant frequency, regardless of the reason for the initial HAZ change. It acts more like a frequency control, rather than a HAZ control. It is difficult to see that a variable frequency welder, with the proposed HAZ control system, gives the tube and pipe manufacturer any real added value in maintaining the HAZ and weld quality for every production batch of a product. In other words: real and true HAZ control requires weld setup control. In a welder with a constant internal inductance (no step- less frequency adjustment), there is no extra adjustable inductance present that can reduce or mask a change in frequency due to a deviating weld process parameter. A repeated and unchanged weld frequency (and power) is then the direct result of a successful reproduction of the reference production weld setup, heat affected zone and weld quality. References [1] “Temperature distribution in the cross-section of the weld Vee”, J.I. Asperheim, B. Grande, L. Markegård, J.E. Buser, P. Lombard, Tube International Nov.1998 [2] “Temperature evaluation of Weld Vee Geometry and Performance”, J.I. Asperheim, B. Grande, Tube International Oct. 2000 [3] “Factors Influencing Heavy Wall Tube Welding”, J.I. Asperheim, B. Grande, Tube International Nov. 1998, Tube & Pipe Technology, March/April 2003 [4] “Selecting a welding frequency”, P. Scott, Tube & Pipe Journal, Oct./Nov. 2003 [5] “System and method of computing the operating parameters of a forge welding machine”, Scott et al United States Patent US 7,683,288 B2 March 2010 [6] “Controlling the Heat Affected Zone (HAZ) in HF Pipe and Tube Welding”, P. Scott, SME March 2007 [7] “HFI Goes Offshore-The Influence of Welding Frequency in Production of Thick-Walled HFI Pipe”, H. Loebbe, Tube & Pipe Technology Sept./Oct. 2005 [8] “A Study of the Key Parameters of High Frequency Welding”, P. Scott, Tube China ’95 Conference, Nov. 1995 [9] “Maximizing Output in High-Frequency Tube & Pipe Welding”, B. Grande, O. Waerstad, Tube & Pipe Technology, March 2012

paragraph is based on an incorrect input to the system. The purpose is to see how the system responds to this flawed input. First, we assume that the vee length is longer than the input value entered into the system. The process response to a longer vee length is a lower frequency. The system is not aware of the wrong input, and the HAZ control concept responds by adjusting frequency up, in order to calculate a HAZ width equal to the one in the reference run. The initial increase in heating of the corners is again reinforced by the increase in frequency. The opposite amplification will take place in case of a vee length shorter than the entered input value to the system. Considerations and limitations The calculation model used in the HAZ control concept presented by Scott and others has two shortcomings [5, 6] : 1. The high frequency current in weld vee is assumed uniformly distributed in the strip wall [8] 2. The proximity effect in the weld vee is not taken into account in the model The first limitation results in a current and temperature distribution in the x-z-plane as shown in Figure 6. This is a 1D model of the HAZ. However, the HAZ is two-dimensional (2D) in the x-z-plane for a large range of wall thicknesses. This is shown in Figure 1, where the hour-glass shape of the HAZ is evident. This implies that the equations used in the 1D calculation model do not accurately describe what happens, electrically and thermally, at the inside and outside corners of the strip edges in the weld vee.

When the proximity effect – a fundamental effect in high frequency current welding – is not a part of the weld vee model, it means that changes in weld vee angle, springback and other geometrical parameters in the weld vee are not properly handled by the proposed concept. A real 2D model (Figure 7) takes into account the proximity effect and can describe the effects that take

Figure 6: 1D model of HAZ

Figure 7: 2D model of HAZ

place at the strip corners and in the tube wall centre when high-frequency current is present. Although pointing out the weld vee angle as one of the parameters affecting the HAZ width [5, 6] , the proximity effect’s influence on the 2D temperature distribution in the x-z-plane of the HAZ is neglected in the proposed system. It can be argued that a 1D model is valid for thin-walled products, where the 2D hour-glass shape of the HAZ is less pronounced. The wall thickness at which a 1D model can replace the real 2D model depends on strip material and weld vee angle. Figure 1a shows the 2D model’s validity for a wall thickness of 2.8mm (0.11"). The hour-glass shape is pronounced in this picture, indicating that the 2D model must be valid for even thinner products. A theoretical study based on Finite Element Analysis shows that the HAZ is still two- dimensional at a wall thickness equal to 1.27mm (0.05") for low-carbon steel [4] .

EFD Induction as – Norway Email: sales@no.efdgroup.net Website: www.efd-induction.com

119

www.read-tpt.com

J uly 2012

A rticle Maximising Output in High-Frequency Tube and

Pipe Welding By Bjørnar Grande and Olav Wærstad, EFD Induction

• Ensure short changeover times, with minimal operator intervention • Contribute to easy start up of new products, with minimal scrap production The topic of achieving consistent high weld quality was covered in the paper Maximising Output in High-Frequency Tube and Pipe Welding 2 . This document is a continuation of that paper, and focuses on how to achieve consistently high weld quality and welder flexibility during changeover. Important features during changeover Minimising scrap requires that the least possible amount of steel strip is consumed during a changeover. This means that the mill and welder parameters from previous successful production runs should be available as a recipe for the next product. The recipe should be downloadable to the welder’s control system, and should be used to automatically preset the required settings for automatic power/ speed control for the product to be welded. These settings are: • Adjustable mill speed at which weld power is turned on, to minimise scrap • Start weld power offset • Weld power-speed gain slope Where temperature monitoring is in use, the weld temperature set-point and acceptable temperature tolerances must be included in the recipe. The use of recipes relieves the operator from performing test runs to find the correct power input and weld quality for the next product. It must be emphasised that successful changeovers do not rely solely on the use of recipes for the welder parameters. Experience shows that the mill (weld) set-up is extremely important for weld quality and power consumption. The mill set-up parameters should definitely be defined in a mill set-up recipe and, together with the welder recipe, should be available for the overall mill quality system. Recipes for existing products can also be used as good starting points when new products are to be welded. This minimises scrap, and reduces start-up times for new products – thereby maximising mill throughput. At the end of changeover it is important that the operator can quickly see whether the mill is properly adjusted or not. He should also see that coil and impeder size and position, as well as impeder cooling, are correct. If these parameters are within reasonable limits and the correct welder recipe is downloaded, energy consumption will be

Abstract The authors evaluate the parameters that influence welder performance and scrap production during changeover in the high- frequency tube and pipe welding process. The paper focuses on the welder system’s features during changeover. The parameters involved are welder recipes, energy consumption monitoring, and matching capabilities. Introduction Maximum throughput in a high-frequency tube and pipe mill is achieved by a welder that features: • High uptime • Consistently high weld quality (to minimise scrap) • Flexibility • High total electrical efficiency High uptime is a prerequisite for high throughput and was addressed in the paper Maximising Uptime in High-Frequency Tube & Pipe Welding 1 . Key design features for maximising uptime are: • The welder must withstand short circuits • The welder must work with high ambient and cooling water temperatures • The welder should not feature continuously operating mechanical parts in order to avoid problems caused by fatigue, wear and jamming • The welder should be based on IGBT transistors, the most rugged inverter switch available Flexibility means a welder that can: • Perform over a wide product (tube/pipe/profile) range • Weld different materials

Figure 1

222

www.read-tpt.com

M arch 2012

A rticle

roughly equal to that achieved in a previous successful production run. This indicates that the set- up is within specified limits. Therefore, the welder system should include the possibility to display the energy consumption factor and the accepted tolerances from the previous reference run. The tolerance values should be available as part of the welder’s recipes.

Figure 2

place more demands on the operator, and require more test runs at changeover, thereby increasing scrap. The best overall solution is a welder with a broad matching range to cope with unexpected operating conditions and the practical tolerances required and given by the total weld process. A welder offering a total operating area, completely without any unsafe areas is, without doubt the best choice (Figure 2c). The EFD Induction Weldac offers this feature, thereby ensuring easy operation during changeover. This in turn minimises scrap and changeover time. Conclusions Maximum throughput in a high-frequency tube and pipe mill requires a welder that contributes to consistent quality and minimum scrap production. The evaluation of the parameters influencing quality and scrap production, conducted both in this and a preceding paper 2 , has led to the following conclusions: • Stable weld temperature requires a weld output power without low frequency ripple. A welder with a passive diode rectifier, some smoothing circuitry and rapid power regulation in the inverter is the best overall solution. This is particularly true in order to meet the strict requirements of high speed mills and mills producing stainless steel tubes. • Recovery after short circuits in the load is optimised by welders with ultra-fast power regulation in the inverter. • The use of welder recipes, including energy consumption monitoring, minimises scrap during changeovers. It also ensures fast changeovers and repeatable quality and production. • EFD Induction strongly recommends welders with automatic matching, without any unsafe or restricted operating areas.

Changeover and welder flexibility The mill operator has many tasks to perform during a changeover. In this situation, it is beneficial if the operator does not have to perform several adjustments to the welder or coil in order to achieve safe and reliable welder operation. Different welder designs influence this part of the operator’s workload. Some welder manufacturers offer welders with matching capabilities, while also offering cheaper versions that lack this important feature. In other words, the welder does not have any means of matching the load to the power supply. Other manufacturers such as EFD Induction ensure that all their welders are equipped with a matching range. A welder with some means of matching – ie featuring a matching range – is a welder with the capability to match the different loads’ electrical characteristics (impedances) to the power supply of the welder, in order to deliver nominal (maximum) output power. The impedance is influenced by the tube dimensions, mill weld setup and induction coil size and position. For a welder without matching range the coils must be specially designed to match the load (coil and steel strip) to the welder’s power supply. These welders have only one operating point at which nominal power is available. It is not feasible to reach this single point for more than a few tube dimensions through coil design alone. This means that the welder’s available output power and weld speed is extremely sensitive to weld set-up variations. If the single optimal coil is damaged, a replacement coil, originally designed for a larger size tube, will reduce available power and throughput. In a situation like this, the number of test runs and the amount of scrap will increase. Moreover, to obtain nominal power through coil design can also lead to a coil that does not optimise the weld process. The end result compromises throughput in steady state operation, not only during changeover. A welder with some means of matching may not be straightforward to use during changeovers. Whether or not this is the case depends on how the matching feature is implemented. Welders are available with and without some matching range, where parts of the total operating area are unsafe (Figure 2a and 2b). In these cases the operator is responsible for running the welder within the safe area. The welder (inverter part) is likely to be damaged if operated in unsafe areas. Welders with such implementation of matching are better than welders without matching range, but they clearly

References

[1] “ Maximising Uptime in High-Frequency Tube & Pipe Welding ”; B Grande, JK Langelid, O Waerstad, Tube & Pipe Technology , March 2011 [2] “ Maximising Output in High-Frequency Tube & Pipe Welding ”; B Grande, O Waerstad, Tube & Pipe Technology , September 2011

EFD Induction as – Norway Fax: +47 35 50 60 10 Email: sales@no.efdgroup.net Website: www.efd-induction.com

223

www.read-tpt.com

M arch 2012

W elding E quipment

Consistent quality in high- frequency tube and pipe welding

will be different, since A1 1/4

is less than A2 1/4 .

At a lower weld speed the heating time is longer. Using 8.25 times the cycle time of the ripple as an example, the difference in total area, due to the difference in A1 1/4 and A2 1/4 , will be almost half the value at the high speed. This shows that the ripple has a larger impact on weld power stability at high speeds than at low speeds. The second situation where the amount of ripple often plays an important part is high-frequency welding of stainless steel tubes. These steel types contain a substantial quantity of chromium that oxidises during welding. The chromium oxide, together with other oxides, forms a hard refractory material with a higher melting point than the base steel. Unless the weld temperature is increased to get molten material across the whole faying surfaces, these solid particles are trapped inside the weld due to poor squeeze out. Conversely, if too much material is melted, the weld vee may become unstable, with possible weld defects as a result. The temperature window when welding stainless steel is, therefore, narrower than for low carbon steel, and a ripple in output power will have a larger effect on weld quality and scrap production. There are three ways to handle the unwanted ripple: install smoothing circuitry (DC capacitor, DC choke or both), regulate power after rectification of the AC mains, or a combination of these two alternatives. The first option is the only one for vacuum tube and solid state welders with a controlled rectifier (SCR). These welders rely solely on installed smoothing and filtering circuitry, which tends to be rather heavy and bulky equipment. Some welder manufacturers have minimised smoothing circuitry, and instead added extra filters in units for stainless steel welding. Maladjustments or control electronics timing problems of the SCR can create non-symmetric stress and reduced service intervals or lifetime of a mains transformer in the factory’s power supply grid. Misfiring of the rectifier’s switches can also lead to a higher ripple at an even lower ripple frequency, thereby increasing the risk of weld quality problems, even at lower weld speeds. It is then a question whether the DC smoothing circuitry is sufficiently dimensioned to cope with such non-ideal

Abstract The authors evaluate the parameters influencing weld quality and scrap production in high-frequency tube and pipe welding. The paper focuses on the welder. Two stages of the production process – steady state operation and non-ideal conditions – are investigated. The parameters involved are ripple in output power and short circuits in the load. Maximum throughput in a high- frequency tube and pipe mill is achieved by a welder that offers high uptime, consistent high-weld quality, flexibility and high total electrical efficiency. High uptime is a prerequisite for high throughput and was addressed in the paper “Maximising uptime in high-frequency tube and pipe welding” 1 . This paper focuses on how to achieve consistent high weld quality. Consistent quality minimises scrap Ripple in the output power is a well-known challenge when trying to obtain consistent welding temperatures. The welder power supply’s rectifier converts the AC mains supply voltage and current to DC voltage and current. This is then fed to the inverter, creating the power supply’s high frequency alternating output voltage and current. The most widely used rectifier types are the diode rectifier and the thyristor controlled rectifier (SCR). Both of these are of the line-commutating type and will, therefore, be the origin of the ripple on the DC voltage and current.

Figure 3: DC voltage during power input to volume ΔV2

Should no action be taken to avoid ripple in the output power, the weld temperature will vary with a stable ripple frequency dictated by the mains frequency. 50 and 60Hz mains supply results in 300 and 360Hz ripple frequency, respectively. The consequences of such a ripple depend heavily on the magnitude of the ripple. There are two situations in which the ripple can negatively impact weld quality. The first is at a high weld speed on small tubes. For weld speeds in the 150-200m/min (~500-650ft/min) range and tube outside diameter in the 12.7-15.9 ( 1 / 2 "- 5 / 8 ") range, and with a distance of around 32mm (1.25") from induction coil to weld point, the heating time of the strip edges will be 9-13ms. This corresponds to 3-4.5 times the cycle time for 300-360Hz ripple. To further describe the situation, we look at two ‘infinitely’ small volumes of material in the strip edges on their way towards the weld point, as shown in Figure 1. The volume ΔV1 enters the weld zone first and the heating time is given by the length Lv and the weld speed. Volume ΔV1 experiences a power that is related to the DC voltage indicated in Figure 2, which shows the non-smoothed DC voltage when using a passive diode or thyristor-controlled rectifier (at full power). Volume ΔV2 enters the weld zone just after volume ΔV1 and will be heated during an equally long heating time as ΔV 1, in this example 4.25 times the cycle time of the ripple. But ΔV2 will face a different power input, indicated by the corresponding DC voltage in Figure 3. Due to the ripple and the different starting point with respect to time, the average voltage (and power), indicated by the shaded areas,

Figure 1: Heating length Lv of volumes

Figure 4: Converter structure, power control in the SCR

Figure 2: DC voltage during power input to volume ΔV1

134

S eptember 2011

www.read-tpt.com

›

W elding E quipment

the welder’s operating point out of the full power matching range. It is essential for reduced scrap production that the welder re-establishes steady-state operation when the short circuit burns off. Fast frequency and output current and power regulation are obvious benefits in this respect. The EFD Induction Weldac has electronic automatic matching and a broad matching range. It rapidly alternates between the different load impedances and quickly reverts to the steady-state point. The output power and current regulation is implemented in the inverter, enabling the market’s fastest regulation and minimising ‘non-welded’ segments (pin holes, etc.) in the final product. Arcing is always a consequence of mechanical irregularities in the strip edges caused by poor slitting and forming, or a too narrow vee angle. In case of severe arcing, actions regarding mill set-up must be taken. Conclusions An evaluation of the parameters influencing quality and scrap production concludes: • Stable weld temperature requires a weld output power without ripple. A welder with a passive diode rectifier, some smoothing circuitry and rapid power regulation in the inverter is the best overall solution. Particularly in order to meet the strict requirements of high speed mills and mills producing stainless steel. • Recovery after short circuits in the load is optimised by welders with ultra-fast power regulation in the inverter. References 1 “ Maximising uptime in high-frequency tube & pipe welding ”; B Grande, JK Langelid, O Waerstad, Tube & Pipe Technology, March 2011 2 N Mohan, WP Robbins, TM Undeland, (1989) Power Electronics: Converters, Applications and Design, John Wiley. By Bjørnar Grande and Olav Wærstad EFD Induction Website: www.efd-induction.com

Figure 5: Converter structure, power control by a DC chopper

operation. It is not possible to remove this problem by power regulation at a later stage in the converter. The second option – to handle the unwanted ripple with regulation only, without any filtering components – is rarely employed. Some energy storage devices to secure energy for regulation are necessary. The smoothing circuitry also has a positive effect on the mains power supply’s power factor 2 . The third alternative with power regulation after the rectifier, together with some smoothing circuitry, is possible in welders where a DC chopper or the inverter takes care of the power control. In this case the rectifier can be of the passive diode rectifier type. The switching frequency of the DC chopper is many times higher than the mains supply frequency, making the chopper response time fast enough for proper regulation. Power control in the inverter, as in the Weldac from EFD Induction, operates directly on the weld frequency level and is the market’s fastest power regulation, reducing ripple to a minimum. Some claim it is best to have a passive smoothing circuit, and not depend on electronics for smoothing the ripple. However, this approach requires attention to

the overall use of regulation circuitry. Solid state welders with controlled rectifier require control electronics for the rectifier and the inverter. Welders with diode rectifier and DC chopper require electronics for controlling the DC chopper and the inverter. Welders with diode rectifier and power control in the inverter require control electronics for the inverter only. The ripple reduction circuitry is an inherent part of the power control for the two latter types of welders. A welder with diode rectifier, some smoothing circuitry and ultra-fast regulation in the inverter, as close to the load as possible, is the best overall solution. The converter topology is shown in Figure 6. As a standard feature, the EFD Induction Weldac guarantees output power ripple to be less than 1 per cent, even with distorted mains supply in the factory. This makes it well suited for high-speed lines and stainless steel welding without the installation of extra smoothing circuitry. Minimising ripple in output power is important for achieving good weld quality during steady-state operation. The loss of a welder’s steady-state operating condition is usually caused by a short circuit in the load. In case of severe arcing, the impedance change of the load can move

Figure 6: Converter structure, power control in the inverter

135

www.read-tpt.com

S eptember 2011

›

A rticle

Maximizing uptime in high-frequency tube and pipe welding Bjørnar Grande, John Kåre Langelid, Olav Wærstad (EFD Induction)

Abstract This article explains some basic principles of solid-state welder design that are crucial for maintaining operation under various conditions. The paper also presents several key differences between MOSFET and IGBT transistors, and describes how a converter with a voltage-fed inverter and series resonant output circuit withstands short circuits. Introduction Tube and pipe manufacturing professionals know the best days are when nothing unexpected happens – when the line works as it should, delivering maximum uptime and throughput. And all of us also realise that the solid-state welder plays an absolutely critical role in achieving and maintaining maximum uptime. Maximum welder uptime requires more than attention to overall circuitry design. Close attention must also be paid to the reliability of each of the components, both in normal and demanding operating conditions. High reliability in steady-state operation is ensured by being in control of the power losses and cooling of the power transfer components. The design must also maintain required margins in relation to maximum component ratings for voltage, current and temperature. Finally, the welder must be able to operate as desired with extremes of water and ambient temperature. For many welders the loss of steady-state operating conditions is usually caused by a short circuit in the load. Arcing can occur between strip edges in the weld vee, between strip and induction coil, or between coil turns or terminals due to slivers and burrs in the weld zone. A welder’s ability to cope with short circuits in the load is, first of all, related to the inverter and the solid-state switches’ short- circuit handling capability. The first part of this paper covers aspects to consider regarding the choice of transistors in the inverter of a welder for high-frequency tube and pipe welding.

Short circuit operation In the tube and pipe industry the output circuit of a welder is available as either a series or parallel resonant circuit. A widely held misconception is that a voltage-fed inverter with a series resonant circuit has inherent problems with short circuits in the load. This misconception stems from the mistaken belief that an arc across the coil causes a flow of high current. On the contrary, what happens is that the resonance point is shifted upwards in frequency. In a series resonant circuit with a high Q-factor, the impedance increases sharply when operating out of resonance and the current drops. [1] The rest of this section explains the events that occur during a short circuit of the series resonant circuit. Figure 1 shows the impedance changes seen from the voltage- fed inverter during a short circuit in the coil. At the instant the arc occurs, the resonance frequency of the output circuit increases and the impedance curve moves up in frequency. The switching frequency of the inverter does not change instantaneously and the inverter will face higher impedance. Arrow one in figure 1 shows this instantaneous increase in the impedance, which results in a current drop from the inverter. Switching frequency increases rapidly towards the new, higher resonance frequency. With the fast current regulation in the inverter, there is enough time for a controlled change of current towards the new operating point, slightly above the new resonance frequency (see arrow two). When the short circuit disappears, there is an instantaneous decrease in resonance frequency and a corresponding increase in impedance, shown by arrow three, followed by the final adjustment back to the previous steady state operating point. No high and dangerous current occurs either in the inverter or elsewhere in the welder due to the short circuit. When a coil short circuit occurs, the load resonant frequency increases. This causes current zero crossing to happen before inverter voltage switching. This type of switching is termed capacitive switching. Using a MOSFET without a series diode considerably raises the risk of activating the MOSFET’s parasitic bipolar transistor (see figure 2). This will immediately destroy the MOSFET transistor. There are ways to prevent this, but they have drawbacks such as startup problems and difficulties recovering from a short circuit during welding. In IGBT transistor modules there are added ultra-fast and soft- recovery freewheeling diodes. These make a short across the coil completely harmless for the IGBT inverter – provided there is a function to limit how long the arc is allowed to burn. Short circuits in the load in tube and pipe welding are in this context very short. In fact, due to EFD Induction’s fast regulation of frequency and current, IGBT transistors even survive long-duration short circuits. A video demonstration of this can be seen on the EFD Induction website [2] .

Figure 1

141

www.read-tpt.com

M arch 2011

›

A rticle

Current sharing between transistors and inverter modules Almost all manufacturers of welding machines today use the principle of a modularised inverter. In order to get the required output power, several inverters are stacked to operate in parallel. Several inverters connected in parallel on both the DC and output sides require stringent control of the turn-on and turn-off of the transistors. The timing in the driver technology is critical, especially in the case of inverters using MOSFET transistors. This is because parameter spreading in MOSFETs is relatively large, which causes variations in the transistors’ turn-off instants among paralleled devices. The slowest transistor to turn off is likely to be destroyed, due to unevenly distributed power loss among the devices. This is the main reason why replacement MOSFETs must be carefully selected prior to installation. It is also the reason why replacement inverter modules for certain MOSFET welders must first be tuned to a specific location in the inverter stack. IGBT transistors, however, can be used off the shelf. There is no time-consuming measuring and pre-selection. This is due to the extremely well proven production process of the non-punch through (NPT) IGBT chips. The production process gives a very tight spread in parameters (such as time delay on/off and gate threshold voltage) compared to epitaxial grown MOSFET transistors. In the EFD Induction welder there are no restrictions on module positioning in the inverter. Position does not affect current distribution among modules, as the overall circuit design guarantees 100% equal current sharing between all inverter modules (as is shown in Figures 3 and 4). There is no need to select driver boards based on time-delay differences. Figure 3 shows the current from two inverter modules, one positioned at the top of the inverter, the other at the bottom. It is difficult to see that these are the current signals from two inverter modules, since they are in fact 100% identical. Figure 4 is therefore the same as Figure 3, but with channel two shifted down one division to show that there are two measured currents. With 100% equal current in all inverter modules – together with the homogeneity of the transistor modules – power loss among inverter modules and operational temperatures of the IGBT transistors are extremely consistent and controlled. Furthermore, at 35°C (95°F) water inlet temperature to the welder, EFD Induction’s design criterion is for a maximum 75°C (167°F) chip temperature inside the IGBT transistor module. The rated maximum chip temperature of the transistor module is 150°C (302°F). The benefit of this system is that both module and system reliability are maintained at the highest level.

Figure 2

switching technologies, EFD Induction’s patented section split system makes the maximum effective switching frequency for one IGBT module of a 400kHz system to be one quarter, that is, 100kHz switching for each IGBT module. This makes the driving of the IGBTs much easier compared to a standard de-rating technique (less driver losses at turn-on and turn-off). A weld frequency of 500kHz with IGBT-based inverter modules is now readily available. The major benefit is the high increase in efficiency compared with a traditional de-rating technique. Based on the same loss level, EFD Induction’s section split system gets 2.5–3 times as much power out of the same IGBT chip area compared to less sophisticated methods. The overall benefit for tube and pipe manufacturers is efficient power transfer at high frequencies with the IGBT transistor’s extremely high reliability. Output circuit A specific weld process – with a specific frequency, coil current, output power and coil – results in a coil voltage that is independent of the brand of welder used. The laws of physics dictate that low internal inductance results in low total voltage. Any added adjustable series inductance (such as for power matching or frequency adjustment) adds extra voltage. As a result, the compensating capacitor voltage installed inside the unit must be higher. The EFD Induction welder is designed with low, and no extra, internal inductance in order to secure low voltage operation. High-power output compensation capacitors are a vital part of a welder. Commercially available capacitor types tend to have either too high internal inductance or a mechanical design which do not take into account the thermal expansion of the capacitor

Figure 3

Figure 4

IGBT at high frequencies

Until EFD Induction introduced its patented switching technique for IGBT transistors, the generally accepted highest frequency range for IGBTs was 125-150kHz. Above this level switching losses became too high without considerable de-rating of output power, making the component uneconomical. Compared with standard, traditional

142

M arch 2011

www.read-tpt.com

›

A rticle

safe operation. No parts of the welder need a dedicated chilled room when operating in high ambient temperatures.

elements during operation. EFD Induction-made capacitors are low-inductance, high-current modules, and are specifically designed for high-frequency welding applications. To ensure long lifetime, two main design criteria are a maximum hot spot temperature of 70°C (158°F) at maximum reactive power, and an allowance for thermal expansion of the capacitor elements. The design is well proven and has been improved and refined over the past 20 years. The numerous internal capacitor elements are double-sided water- cooled in order to secure high current and high reactive power operation. Flow switches monitor the water flow, and each capacitor module has a dedicated thermostat for additional protection. Some welders with a parallel resonant output circuit use variable series inductance as a way of obtaining some matching capabilities. The major disadvantage of this solution is the number of moving electromechanical parts in the output circuit, which are prone to wear and jamming. Should relays and electrical motors be used for controlling, these components are also likely to face fatigue problems. The EFD Induction Weldac has automatic electronic load matching in the inverter which does not require the continuous operation of moving mechanical components [3] . No extra mains transformer maintenance In a welder with no intermediate transformer, the output circuit is not isolated from the mains supply. A mains transformer is then required, either installed outside or inside of the DC power supply cabinet. Where transistors with low-breakdown voltage are used in the inverter, a transformer is also needed to adapt to the factory’s higher mains supply voltage. It is especially critical that welders with power control in the SCR have very precise control and firing of the thyristors in the rectifier. This is to avoid non-symmetric load of the three windings in the mains transformer. Incorrect adjustments and timing differences result in non-symmetric stress, which reduces maintenance intervals and/or lifetimes of mains transformers. The EFD Induction welder includes an intermediate, low-loss transformer for both loadmatching and galvanic separation purposes. A mains transformer to insulate the output circuit from the mains is not required. Because of power control in the inverter, the EFD Induction welder uses a passive diode rectifier. This does not cause any non-symmetric load or stress on any mains transformer in the tube manufacturer’s factory power supply grid, further enhancing uptime. The output compensating capacitors in the output circuit are on the secondary side of the intermediate transformer. Due to this, no reactive power transfer takes place in the transformer and a low voltage operation is secured. The windings and core are moulded in a resin without any oil content. Water & ambient temperature The EFD Induction welder is designed to operate at ambient temperatures of 5° to 50°C (41° to 122°F). All power components inside the cabinet(s) are water cooled. The water-cooling circuits are designed for a water inlet temperature of 35°C (95°F), and flow is monitored by flow switches. Several components are additionally protected by thermostats. Furthermore, a water/air cooler is installed inside the cabinet(s) to keep inside ambient temperature within the range for all components, including the electronics. The cooling water temperature is controlled by the water/water-cooling system to keep the water temperature inside the cabinet above the dew point. Where necessary, an air conditioning unit is included for extra

Summary A successful welder design for high-frequency tube and pipe welding must maximise uptime and throughput. To achieve this objective in the relatively harsh environment of a tube mill, EFD Induction has designed the Weldac. The following were key design objectives: • The welder must be able to withstand short circuits • The welder must work with high ambient and cooling water temperatures (caused for example by climate conditions) • The welder must operate with the lowest possible voltage in the output circuit • The welder should not feature continuously operating mechanical parts (in order to avoid problems caused by fatigue, wear and jamming) • The welder should feature readily available components (such as ‘off-the-shelf’ IGBTs). The Weldac is based on a voltage-fed inverter and a series resonant output circuit, and easily handles short circuits. No large currents or overvoltages occur during short circuits. This type of welder operates safely and reliably over a very wide frequency range. EFD Induction has 30 years’ experience with solid-state switches in the inverters of induction heating equipment. During the last 20 years, EFD Induction has gained extensive experience with both MOSFET and IGBT transistors in high-frequency tube and pipe welding. Where consistently high uptime and output are priorities, the IGBT transistor is the inverter switch of choice: • The IGBT has an intrinsic short-circuit handling capability: it is an extremely rugged component • Because of tight parameter spreading, the IGBT is the best choice for paralleling of transistors. Combined with the patented section split system (which improves current sharing among paralleled transistors and reduces the required number of paralleled transistors) this gives a very reliable system • The IGBT is a widely available and standard industrial transistor. Unlike MOSFET welders, there is no need to carefully select and tune transistors and inverter modules. The overall benefit for a tube and pipe manufacturer is efficient power transfer at high frequencies (70-500kHz) with IGBT transistors’ extremely high reliability. One consequence of this rugged design is that EFD Induction is the only tube and pipe welder manufacturer to offer a five-year warranty for the system’s inverter modules and driver cards.

References

1 N Mohan, WP Robbins, TM Undeland, (1989) Power Electronics: Converters, Applications and Design, John Wiley. 2 www.efd-induction.com/en/bestwelder 3 F Kleveland, JK Langelid, L Markegård, (2003) “New HF Converter for Induction Heating”, Proceedings of the International Conference on Electromagnetic Processing of Materials, Paris.

EFD Induction PO Box 363, N-3701 Skien, Norway

Tel: +47 35 50 60 00 Fax: +47 35 50 60 10

143

www.read-tpt.com

M arch 2011

›