Parameter optimization of AC rectangular wave outputs for aluminum cold wire GTAW

Parameter Optimization of AC Rectangular Wave Outputs for Aluminum Cold Wire GTAW

Américo Scotti

Roberto Aparecido dos Reis

Universidade Federal de Uberlândia

30400-902 Uberlândia, MG Brazil

[e-mail]

Abstract

The objective of this work was to optimize the parameter setup for GTAW of aluminum using an AC rectangular wave output and continuous feeding. A series of welds was carried-out in an industrial joint, with variation of the negative and positive current amplitude, the negative and positive duration time, the travel speed and the feeding speed. Another series was carried out to investigate the isolate effect of the negative duration time and travel speed. Bead geometry aspects were assessed, such as reinforcement, penetration, incomplete fusion and joint wall bridging. The results showed that currents at both polarities are remarkably more significant than the respective duration times. It was also shown that there is a straight relationship between welding speed and feeding speed and this relationship must be followed for obtaining sound beads. A very short positive duration time is enough for arc stability achievement and when the negative duration time is longer than 5 ms its effect on geometry appears. The possibility of optimizing the parameter selection, despite the high inter-correlation amongst them, was demonstrate through a computer program. An approach to reduce the number of variables in this process is also presented.

Keywords: AC GTAW, Square wave GTAW; Aluminum; Parameter Optimization; Experimental Design

Introduction

Welding of aluminum and its alloys presents some peculiarities in contrast to ferrous materials, due to the physical and chemical properties of aluminum (passive oxide layer, high thermal and electric conductivity, low fusion temperature). Every welding process (SMAW, GMAW, GTAW, PAW, etc.) exhibits advantages and limitations for aluminum welding. The right process is chosen according to the required quality and productivity (and investment capability) for a given joint and material application. GTAW is a reasonable process to minimize the operational problems of aluminum welding. This process provides quality welds with tolerably low distortion. It can be used in any position and requires no flux. It is especially applicable to thin-walled components (less than 5 mm - ³/₁₆ in.) and/or where high production rates are not required, yet the GTAW process has had a substantial increase in productivity by using continuous welding wire feeding, specially in small and medium plants (Richardson, 1995).

From a process point of view, the main problem is arc instability and consequent porosity. A power source with DC output and electrode negative (DCEN) demands special care for aluminum welding in comparison with steels. Despite an efficient thermoionic electron emission, which is theorized to predominate in DCEN, aluminum alloys form refractory surface oxides. As the oxide layer has a fusion temperature much higher than that of the metal, heat penetrates the oxide and can melt the base metal, but not melt the oxide, which acts as a "stop-off" to prevent fusion of a weld. This oxide layer also leads GTAW with electrode negative to porosity (AWS, 1991). According to Barhorst (1985), DCEN GTAW can work only if the aluminum is extremely clean prior the welding (chemical etch followed by manual scraping). Helium or helium-argon mixtures with at least 65% He must be used instead of argon as shielding gas (Anik and Dorn, 1992).

Porosity in aluminum is caused mainly by hydrogen, which in turn arises from the arc (instability or shielding gas contamination), from the welding wire or from the surface oxides (Martukanitz and Michnuk, 1981). Even a plate pre-cleaning (chemical and/or mechanical), which use is recommended in practice (Norrish and Ooi, 1993), mainly concerning oily substances, is able to prevent porosity (Barhorst, 1985, and Tomsic, 1984). It is believed that an excessively clean surface may result in poor stability (Norrish and Ooi, 1993), specially using power supplies with low open circuit voltage (OCV), but some types of aluminum may demand this cleaning (Tomsic, 1984). Roughening the "clean" surface with a wire brush is advisable by manufactures to stabilize the arc with low OCV machines.

A stable arc can be established without the natural oxide removal, but not with water stained or anodized surfaces with thicker oxides, which become electrical insulators.

On the other hand, when the electrode is positive (DCEP), a workpiece oxide self-cleaning phenomenon is assumed to take place, so that a stable arc could be established (the aluminum plate becomes the electron emitter). The phenomenon of oxide self-cleaning is explained on the basis of the electron emission:

Electron emission from non-thermionic materials (such as aluminum) is only possible by the phenomenon of "field emission" (Lancaster, 1987). Field emission is a resultant of a very high voltage gradient between a metal cathode (negative pole) and a positively charged oxide layer. The existence of this positively charged oxide layer is due to incident positive ions onto the surface oxides, which are formed by reaction between oxygen and the workpiece elements. The oxides are believed to act as a source of electrons because they usually have a lower work function than the correspondent metals (Jönsson, Murphy and Szekely, 1995). It appears that this mentioned high voltage gradient happens because the electrons concentrate in some sites (spots). This locally generated temperature is enough for melting or evaporating the oxides, which, by surface tension or vaporization, are removed from the weld pool surface, causing the oxide self-cleaning.

There are other theories by means of which this cleaning effect can be explained (Anik and Dorn, 1992). In a first one, the electrons leaving the workpiece at high velocity tear off the oxide skin and break it down into very small particles. In another one, the ions striking the workpiece have sufficient energy to shatter the oxide skin. Evidence for this theory would be the fact that the cleaning effect with an inert gas of high relative atomic mass (argon) is more pronounced than with a lighter one (helium).

With two thirds of the heat energy generated at the anode (positive pole) and one third at the cathode (negative pole), the DCEP weld pool is wide and shallow (Barhorst, 1985) and the electrode experiments a high wear rate (Dutra et al, 1991). DCEP aluminum welding is also prone to porosity and oxide inclusion (Tomsic, 1984) and limited to only the very low amperage (thin materials) applications (Barhorst, 1985).

Considering the limitation of using DCEP, the problem due to the oxide layer has been overcome by the use of sinusoidal AC output signal overlapped by a high voltage-high frequency signal (HF). A higher open circuit voltage of the transformer can be also demanded. During the positive half cycle, the oxides on the pool are cleaned, while the weld is performed mainly during the negative half cycle. When the electrode becomes negative, it supplies electrons immediately to restart the arc (thermionic emission). However, when the weld pool becomes negative, it cannot supply electrons until voltage is raised sufficiently to initiate cold-cathode emission (AWS, 1991).The HF signal and the high open circuit voltage have the function of assisting the arc stability. One interesting fact is that, even using an overlapping HF signal, a voltage and current rectification effect take place during the positive half cycle (I- > I+). Such rectification can cause damage to the power supply due to overheating or, with some machines, a decay in the output (AWS, 1991). Dutra et al. (1991) demonstrated that using capacitors in series to balance the wave could eliminate this effect. The capacitors are charged during the negative half cycle and discharged during the positive half charge. However, they did not find any significant improvement in the weld and the electrode wearing was increased. Other source (AWS, 1991) also points out for a need of larger electrodes if balanced wave is employed, though a better oxide removal and a smoother welding are claimed.

However, when automation of the process is targeted, the HF signal may become a drawback due to the interference in the electronic devices (Barhorst, 1995, Martukanitz and Michnuk, 1981 and Tomsic, 1984). This interference happens even if they are positioned far from the power source, since the HF signal emits in radio-frequency, through electrical wiring or air (AWS, 1991). One solution for this problem is to use the so-called square-wave AC output signal. The advantages of using this technique instead of the conventional sine wave are widely known (see for instance, AWS, 1991). The first purpose of this approach is the reduction of the transition time between polarities; if this time is too long (as in sinusoidal output signal), the arc voltage and current might reach values insufficient to maintain the required arc stability. At a rapid switching rate, the arc remains ionized through the transition times. Norrish and Ooi (1993) cites that the use of high frequency transistor reversing switch circuits enable a "square wave" ac output to be obtained, with capability of reversing the polarity of the output within 100 ms. They also mention that the reignition process during polarity reversal may be further assisted by arranging for high voltage transients to coincide with current zero. This approach is essential when current reverses from negative to positive polarity (Lin, Wang and Zhang, 1989).

A second purpose of using square wave signal is to increase the thermal efficiency of the process, by lengthening the periods with negative polarity (t-) at the expense of the periods with positive polarity (t+). Accordingly, the positive polarity period is set to be just enough for oxide cleaning, leaving longer time for the better heat transfer achieved at the negative polarity cycle. This approach is claimed to increase metal fusion (penetration) and to reduce electrode wearing, facts that improve the productivity of the process. More precisely defined energy input is also expected, as demanded to weld thin sheets, as stated by Rehfeldt et al, 1996. As can be seem, the correct denomination for the process should be "Asymmetric Rectangular Wave AC GTAW", instead of the more used term square wave, as illustrated in Fig. 1.

Fig. 1

However, despite all knowledge found in the current literature, there is very little information concerning the optimum parameter settings. Some data are even contradictory. Tomsic et al. (1984) report that, for Plasma Arc Welding (PAW), the level of positive polarity current must be higher than the negative one to get oxide cleaning. The most important is the cycle, in which t+ must be in between 2 and 5 ms and t- within 15 to 20 ms. Barhorst (1985) worked with 8 ms of DCEN and DCEP polarity time in some of its experiments. Dutra et al. (1992) obtained good results with a t+ equals 4 ms, t- equals 12.6 ms and a rms current of 73 A. Anik and Dorn (1992) suggest in their work a positive duration time as short as 0.2 ms and a negative duration time of 2.5 to 10 ms.

Norrish and Ooi's (1993) found that only a limited GTAW operation range, with an AC balance around 50%, was usable if acceptable arc stability was to be achieved. With a DOC system (Dynamic Oxide Control) switched on, a range of t- from 25% to more than 75% gave good arc stability and the stability was especially good when plate heating was maximized (i.e., at 75 to 90% electrode negative). DOC was defined as a system in which the t+ period is suppressed if during each polarity reversing the arc is still stable, that is, oxide cleaning is not necessary. The not defined situation concerning parameters setting is further obscure when cold wire feeding is a part of the process or when a joint configuration is applied.

Nomenclature

A/D = analogic/digital conversion

AC = alternate current

Ar = Argon

ASTM = American Society for Testing and Materials

AWS = American Welding Society

CD = corner discontinuity

CON = convexity

DC = direct current

DCEN = direct current electrode negative

DCEP = direct current electrode positive

DOC = Dynamic Oxide Control

GMAW = a process denoted by Gas Metal Arc Welding

GTAW = a process denoted by Gas Tungsten Arc Welding

He = Helium

HF = high frequency

I- = mean current at negative polarity

I+ = mean current at positive polarity

LIF = lateral incomplete fusion

NP = negative penetration

OCV = open circuit voltage

O_p = bead overlapping on the plate side

Os = bead overlapping on the shape side

P_abs = absolute penetration

PAW = a process denoted by Plasma Arc Welding

PP = positive penetration

Re = bead reinforcement (bead height)

SMAW = a process denoted by Shielded Metal Arc Welding

t- = duration time at negative polarity

t+ = duration time at positive polarity

TS = travel speed

W = bead width

WFS = wire feed speed

This lack of more information prevents the users from taking the most advantage of the process. Another limitation would be that different types of aluminum may demands different settings, mainly concerned to positive and negative periods. Therefore, the purpose of this work is to determine an operational envelope for GTAW of aluminum in a groove (joint preparation) with AC asymmetric rectangular waveform and continuous feeding. The work also aims to study the influence of each parameter on the phenomenon of arc stability and bead formation. A parameter optimization approach is eventually expected to be achieved.

Experimental Methodology

In a first moment, six welding variables were chosen to have their influence on the bead geometry studied. After a series of pre-tests to find some feasible parameter combination, an experimental design (half-factorial design) was elaborated with 6 factors (welding variables) at two levels. The lower and higher limit levels (setting values) for each factor were: negative (I-) and positive (I+) current amplitude = 140 and 190 A; positive current duration time (t+) = 1.5 and 2.0 ms; negative current duration time (t-) = 3.0 and 6.0 ms; wire feed speed (WFS) = 25 and 58 mm/s; travel speed (TS) = 2.5 and 5.8 mm/s.

Following the experimental design, 32 test plates were welded in a random order, in flat position, using Ar as shielding gas at a rate of 15 l/min, a 3.2 mm AWS EWP electrode and a 1.2 mm ER5356 (AWS A5.10-88 standard) welding wire. The test plates (Fig. 2) were composed of a 6061-T6 pre-molded aluminum shape and a 5032-H32 aluminum plate, both manufactured by Alcoa in Brazil. Metallic spacers maintained the 1.5-mm joint spacing. This joint configuration is used in truck bodies frames. The welds were carried out using an automatic torch carrier and the arc length was maintained constant at 3 mm from the plate surface along all welds. A water-cooled torch was positioned normal to the test plate surface and the welding wire feeding (cold wire) was made through an appropriate welding wire guide attached to the torch (front feeding into the pool).

Fig. 2 Test plate form and dimensions (units in mm)

For these tests, an electronically controlled (secondary chopped) multiple-process welding power source was used. One special characteristic of this power source, concerning this work, is the capability of individual settings for the AC waveform parameters, i.e., I+, I-, t+ and t- (Fig. 1). This power source provides also a very short 600V capacitor discharge every time the current goes from the negative to the positive period.

Arc voltage and current signals were recorded directly into a microcomputer data acquisition board (A/D). The signal acquisition was made at a rate of 16 kHz at each channel, at 8 bit, for 8 s. Arc voltage was measured between the workpiece and the contact-tip of the torch. A "hall effect" probe in series with the arc enabled accurate monitoring of the current waveform. A visualization software (including zoom capability) permitted to analyze signals (V versus t and I versus t) after welding. The mean values of the amplitudes and periods of current and voltage were obtained by direct measurement on the signal traces (from 6 sampling cycles). The results are presented as the average value of the samples. Wire feed speed (WFS) was measured using a photo-encoder based sensor and travel speed (TS) was monitored using a calibrated feedback signal from the carrier motor. The readings of WFS and TS were made direct from the device displays.

Table 1 presents the mean values of the monitored parameters, according to the settings specified in experimental design. Some of the parameter combinations produced some beads with no bridging on one or both lateral walls of the joint, or whose pool was far high to touch the electrode tip. These misshapen beads were results of a lack of compatibility between parameters, such as high TS with low WFS or vice-verse. To unbalance the experimental design as less as possible, by not excluding these test plates, new tests were implemented with the smallest variation possible in TS and WFS. The new welding parameter combinations that replaced the undesired combinations are presented in Table 2.

Table 1

Welding parameters mean values (monitored) using in the first experimental procedure

Table 2 Welding parameters mean values (monitored) of the replacing test plates

Table 2	Welding parameters mean values (monitored) of the replacing test plates

The acceptability criterion for bead shape was based on the geometric parameters illustrated in Fig. 3. It is important to point out that, as aluminum welding is very sensitive to small variations in plate thickness, some geometric parameters were denoted accordingly the plate side and the pre-molded shape side. The line L₁ represents the space plane of the step in the pre-molded shape. The main defined parameters (indexes) were: NP (negative penetration) - the distance from L₁ to the bottom of the beads which did not reached the pre-formed shape; PP (positive penetration) - the distance that a bead penetrated beyond L₁; P_abs (absolute penetration) - either NP or PP, since one excludes the other; CON (convexity) - represented by the ratio Re/W, where Re is the bead reinforcement (bead height) and W the bead width; O_p (bead overlapping on the plate side) and Os (bead overlapping on the shape side) - bead length over the base metal with no fusion; CD (corner discontinuity) - area at the joint corner not filled by the weld pool; LIF (lateral incomplete fusion) - linear length from L₁ to the point where the bead begins to fuse the joint wall on the pre-molded shape side.

Fig. 3 Geometric parameters to define the bead characteristics (criteria)

The measures were taken on transverse sections of the test plate (2 per test plate). After metallographic preparing and etching (Tucker's reagent), the geometric parameters were measured using a computerized system (digital image analyzer), except the dimensions Re and W, which were obtained by vernier caliper measurement along the bead (16 samples per test plate).

Results and Discussion

Table 3 presents the responses concerning the bead geometric parameters from the first series of experiments. A statistical technique denominated Partial Correlation Analysis was applied on these data to identify the level and significance of the correlation between the welding parameters (factors) and the bead geometric parameters (responses). Table 4 shows the outcome of this approach. A significant correlation is indicated by a correlation coefficient close to 1 (a coefficient of Partial Correlation Analysis quantifies the relationship between one factor and one response taking into account the effect of the other factors on this response). A negative sign before the coefficient means that the correlation is inversely proportional.

Table 3

Mean values of the bead geometric parameters from the first series of experiment (Tables 1 and 2)

Table 4

	Coefficients of the Partial Correlation Analysis between factors and responses from the first series of experiments (Tables 1 and 2)

As seen in Table 4, I+ and I- are the governing factors on penetration (P_abs), but with low significance (I- is slightly more influent). t-, t+, TS and WFS show to have low influence on penetration. Considering that a sound geometric bead is also representative of stable arc, the results showed that the chosen range for t+ was very adequate. One should expect that the factors that affect the penetration should be the same as those that affect the incomplete fusion discontinuities (expressed by LIF and CD). However, the correlation analysis shows an increase of significance of the I- and WFS on the LIF index, while TS appears to be the only significant factor on the CD index. These results suggest that other factors not considered in this analysis, such as groove shape, may play an important role on those geometric parameters, and that the significance of the role varies according to the geometric parameter.

Regarding the convexity (CON), the governing factors are clearer, with good significance for WFS, I- and TS. These more predictable results support the above mentioned suggestion, since groove shape is not expected to affect the bead finish as much as it affects penetration. A greater value of material deposition (high WFS and/or low TS) at the same current is prone to a more convex bead finish. The increase of current (in the case of the GTAW process, I-, and indirectly t-, are more effective than I+ and t+) is inclined to increase the bead wetting (less convexity). It is important to recall that the CON index was defined as a ratio of Re and W. Thus, the influence of the factors on CON, Re and W should be alike. Again, each factor seems to play distinct role on different geometric parameters. As seen in Table 4, the WFS lost the significance on the W, while I+ had its significance remarkably increased. This result appears also to be coherent, since the increase of the deposited material (high WFS) does not mean a correspondent increase of the wetting between the pool and the base metal. This is true in the GTAW process, in which one can increase the WFS and keep the same heat input. On the other hand, an increasing I+ is likely to increase the wetting by enlarging the cathode cleaned area.

Regardless a low significance, the bead width (W) seemed to be controlled by the t- in an inverse direction. Norrish and Ooi (1993) had also found this unexpected result. Finally, from the above, it seems to be justified the major effects of the WFS and TS factors on the overlapping discontinuities (O_p and O_s), followed by the I-.

Another statistical technique was applied on the data from Table 3 with the objective of concentrating the influence of all the factors on the bead formation as a whole. An acceptable bead would have a not very convex finish (CON) and would not present incomplete fusion discontinuities (LIF or DC). The reinforcement (Re) could be neither very high (the electrode tip would touch the pool), nor very low (risk of lack of joint wall bridging); the control of Re is a way of finding the feasible combinations of TS and WFS. Using Multiple Regression Analysis, these most important geometric responses were modeled in relation to the input factors, as seen in the following Eq. (1) to (3). The statistical indexes to assess the adequacy of the models are the "F-ratio number", "P-Value", the correlation coefficient R² and the standard error s. The higher the F-ratio number and the lower the P-Value, more representative is the model regarding the phenomenon. The correlation coefficient R² indicate the fitness of the model to the data with which the model was generated; the closer to the unit the better. The standard error s gives the idea of the predicted deviation between the calculated and actual value.

CON = 0.384 x WFS - 0.0014 (I- x WFS) - 0.00016 x TS² (Eq. 1)

with F-ratio = 200; P-value = 0.0000; R² = 0.9508 and s = 0.07175;

LIF = 0.000049 x (TS x I-) - 3.1037 ÷ WFS - 0.000018 x I+² + 548.133

÷ I- + 0.05534 (t+ x t-)

(2)

with F-ratio = 182; P-value = 0.0000; R² = 0.9712 and s = 0.4401;

Re = 1.0698 x WFS - 0.001589 x TS²

(3)

with F-ratio = 514; P-value = 0.0000; R² = 0.9717 and s = 0.3270.

In order to determine the possible welding parameter (factor) combinations, that is, the combinations which give sound welds from a bead geometry point of view, restriction values (constraints) must be established to each response (geometric parameters), taking into account the limits of the factors used in this experiment. By doing so, a multiple dimension operational envelope (tolerance box) is delineated. The use of computational methods permits solving those equations, and optimized combinations can be found. For instance, if restrictions are arbitrarily imposed to Re (1 £ Re £ 3 mm; 0.04 £ Re £ 0.12 in.), LIF (LIF £ 1mm; £ 0.04 in) and CON (0.09 £ CON £ 0.66), the computational program would provide an optimum combination from the perspective of the travel speed, for the given joint and process, as follows: I+ =184 A, I- = 186 A, t+ = 1.67 ms; t- = 6.01 ms; WFS = 28.3 mm/s (67 in./min); TS = 3.5 mm/s (8.3 in./min). If the allowable value for LIF were more restrictive, such as LIF = 0, the optimum TS would be around the lower value of the studied range (2.5 mm/s - 5.9 in./min). Analogously, an erroneous definition of the constraints (higher permissible Re and LIF values) would provide another combination such as I+ = 184 A; I- = -160 A; t+ = 1.8 ms; t- = 5.9 ms; WFS = 58 mm/s (138 in./min) and TS = 3.5 mm/s (8.3 in./min). These two combinations were used in real welds for validation of the model. Table 5 presents the output of these welds. As seen in this table, there is a good agreement between the expected results and the actual ones.

Table 5

	Comparison between the predicted and actual outputs of the modeling validation

In this new experimental planning, the same joint configuration and electrode and shielding specifications were used. I+, I- and t+ were kept constants along all welds, as a way to isolate the effects of t- and TS. The highest nominal values for currents (190 and -190 A) were employed since they showed to be necessary to get sound welds. The nominal value for t+ was chosen as 2 ms. Four levels of t- (5, 10, 15 and 20 ms) were picked to guarantee a better phenomenon representation. Three levels of TS were elected (2.5, 4.2 and 5.8 mm/s - 9.8, 5.9 and 13.8 in./min). However, the approach for isolating the TS effect was more elaborated. As it is well known, penetration has not a direct correlation with TS. It is dependent on the size of the pool (deposition rate) and current, presenting a maximal at the point where the arc stands on the pool leading root. Fortunately, one can use the WFS independently of de TS in GTAW. Thus, for each level of TS, a proper value of WFS was selected, so that the pool volume was always the same.

The correct value of the pool volume was taken from the optimum combination obtained from the first series of experiment (I+_opt =190 A; I- _opt = 190 A; WFS _opt = 28.3 mm/s - 67 in./min and TS _opt = 3.5 mm/s - 8.3 in./min) and using the following expression (Eq. (4)):

WFS_new/TS_new = WFS _opt/TS _opt = 8.086

(4)

where the subscript new denotes the parameters for the second series of experiment and the subscript opt means the optimized parameters from the first series of experiment (the resultant nominal values for the pairs WFS_new/TS_new were 20/2.5, 33.3/4.2 and 46.7/5.8).

A full 3¹ x 4¹ experimental design was prepared. Table 6 shows the welding parameter combinations used in each test plate (carried out in a random order), whereas Table 7 presents the responses concerning the bead geometric parameters. Table 8 exhibits the outcome from the Partial Correlation Analysis.

Table 6

Welding parameters mean values (monitored) using in the second experimental procedure

Table 7

	Mean values of the bead geometric parameters from the second series of experiment

Table 8

	Coefficients of the Partial Correlation Analysis between factors and responses from the second series of experiment

These results suggested that both t- and TS effects were effectively isolated. The increase of plate fusion related parameters (P_abs, LIF, CD and OS) and the small influence on the geometric parameter W seems to be coherent with the extension of the time under action of the I-. Concerning the TS, the increased heat input effectiveness (base metal penetration) due to lower travel speeds is no longer off set by the pool size.

Two important gains of the TS and t- effect isolation was the possibility of reducing the number of variables in this process and optimizing their settings, some of the main disadvantages of the process. The operator needs only to set the current (I- = I+), t- and TS, as long as he/she knows that t+ must be around 2 ms and WFS is a consequence of a given TS. If one wants, e.g., the highest productivity, current should be set at the most level (according to the electrode size). TS and t- would be set according to the desired geometry (penetration, convexity, incomplete fusion, etc.). One example would be to take the test plate 42 as the best combination (I+= 184 A, I- = 186 A, t+ = 2.2 ms, t- = 10.6 ms, TS = 5.8 mm/s - 13.8 in./min and WFS = 48.3 mm/s - 114 in./min) rather than the optimum combination found in the first part of the work. Another achievement is the possibility of using this reasoning to control the process concerning bead geometry and metallurgical aspects (heat input). Further work is in development, whose main point is the influence of parameters on arc stability, aiming a fully automated system. Both oxide self-cleaning and TS/WFS relationship control strategies are planned to be implemented in an electronic welding machine.

Conclusion

The results of this work showed that current (mainly negative electrode polarity), travel speed and wire feed speed have the mandatory action on weld geometry formation during GTAW of aluminum using rectangular wave AC output signals in the specific grooved joint. The higher the current amplitude, within the electrode permissible range, the deeper the penetration (higher productivity). I or TS, depending on other external factors (such as joint configuration, position of the discontinuities, etc.) can also correct incomplete fusion discontinuities. Parameters producing complete joint penetration does not necessarily guarantee a weld free of incomplete fusion. A strict balance between TS and WFS must be reached to prevent misshapen beads and to reduce the number of variables in the process setup. The augmentation of productivity by TS increasing takes into account the correspondent increase of WFS at a proper rate. The time duration of positive electrode polarity (t+) seems not to affect the weld geometry and a short time (1.5 to 2.1 ms) is enough for cathodic cleaning and arc stability, for those types of aluminum. The effect of the time duration of negative electrode polarity (t-) is dependent on its range. For short values (cycle frequency around 125-220 Hz), t- has minor effect on the geometry. For longer times (cycle frequency around 45-150 Hz), increased t- causes deeper penetration.

Acknowledgments

The authors would like to express their gratitude to BINZEL do Brasil and ALCOA-CARGOVAN for the material and technical support. Thanks are also given to the Universidade Federal de Uberlândia and a Brazilian council for personnel improvement (CAPES) for the financial support.

Manuscript received: February 1999, Technical Editor: Allison Rocha Machado.

• Anik, S. and Dorn, L., 1992, "Metal-Physical Process Involved in Welding – Welding of Aluminium Materials", Welding Research Abroad, vol XXXVIII (3), Mar 1992, pp. 36-41;
• Aws, 1991, "Welding Handbook - Vol II: Welding Process", Chapter 3, AWS, 8^a ed., pp. 74-107;
• Barhorst, S., 1985, "The Cathodic Etching Technique for Automated Aluminum Tube Welding", Welding Journal, AWS, May 1985, pp. 28-31
• Dutra, J.C. et al., 1991, "Estabilidade do Arco TIG em Corrente Alternada", XVII Enc. Nac. Tecnologia da Soldagem, ABS, Recife, Set 1991, pp. 577-589;
• Dutra, J.C. et al., 1992, "Corrente Alternada de Onda Retangular: uma Inovação com Tecnologia Nacional para a Soldagem a Arco do Alumínio", IIW Latin American Regional Welding Congress - 1992, ABS/IIW/Senai, Rio de Janeiro, Abril 1992, pp. 989-1000;
• Jönsson, P.G., Murphy, A.B. and Szekely, J., 1995, "The Influence of Oxygen Additions on Argon-Shielded GMAW Processes", Welding Journal, Feb 1995, pp. 48s-58s;
• Lancaster, J.P., 1987, "The Physics of Fusion Welding - Part 1: The Electric Arc in Welding, IEE Proc", 134 (5), 1987, pp. 233-252;
• Lin, Y., Wang, Q. and Zhang, J., 1989, "Mechanism of AC Welding Arc and Its Reignition, 2nd Int. Conf. on Recent Trends in Welding Science and Technology", ASM, Gatlinburg, USA, 14-18 May 1989, pp.511-515;
• Martukanitz, R.P. and Michnuk,P.R., 1981, "Source of Porosity in Gas Metal Arc Welding of Aluminum, Conf. On Trends in Welding Research in The United States", ASM, New Orleans, USA, 16-18 Nov 1981, pp 315-330
• Norrish, J and Ooi, C.L., 1993, "Adaptive Asymmetric Wave form Control in Bipolar GTAW of Aluminium", Welding and Metal Fabrication, Jun 1993, pp. 230-232.;
• Rehfeldt, D.; Schimitz, T.; Mecke, H. ; Döbbelin and Heyder, D., 1996, "Studies into the Use of Higher Frequencies for Tungsten-inert Gas Welding with Low Alternating Currents", Welding and Cutting, 5, 1996, pp. E90-E93;
• Richardson, M., 1995, "Semiautomatic Cold Wire Feeder Systems Increase GTA Productivity", Welding Journal, Jan 1995, pp. 51-54;
• Tomsic, M.J. et al., 1984, "Welding of Aluminum with Variable Polarity Power", IIW doc no. XII-839-84, 17 pp.

Publication Dates

• Publication in this collection
  11 Oct 2001
• Date of issue
  Dec 1999

History

• Received
  Feb 1999