Abstract

Welding current effect on diffusible hydrogen content in flux cored arc weld metal

A. Q. Bracarense^I; R. Lacerda de Souza^II; M. C. M. de Souza Costa^II; P. E. Faria^III; S. Liu^IV

^I queiroz@vesper.demec.ufmg.br

^II celeste@demec.ufmg.br

^IIIFederal University of Minas Gerais Mechanical Engineering Department Group of Robotics, Welding and Simulation Av. Antônio Carlos 6627, Campus Universitário 31270-901 Belo Horizonte, MG. Brazil

^IVColorado School of Mines Center for Welding, Joining and Coatings Research Golden, Co, 80401, U.S.A. sliu@mines.edu

ABSTRACT

The application of flux cored arc welding (FCAW) has increased in manufacturing and fabrication. Even though FCAW is well known for its good capability in producing quality welds, few reports have been published on the cause of the relatively high diffusible hydrogen content in the weld metal and its relation with the ingredients used in the wire production and with the welding parameters (mainly welding current). This paper describes experiments where data obtained from weld metal diffusible hydrogen analysis, metal droplet collection, and high-speed recording of metal droplet transfer were used to evaluate the effect of welding current on diffusible hydrogen content in the weld metal. The results from gas chromatography analysis showed that weld metal hydrogen content indeed increased with welding current. A polynomial regressional analysis concluded that hydrogen increase with current was better described by a linear function with proportional constant of approximately 0.7 or 70%. Different from the GMA welding transfer behavior, statistical analysis showed only a small increase in metal droplet size with increasing current. The metal transfer mode remained in the globular range for currents between 100 and 150 A. The most surprising findings were with the high-speed cinematography recording. Observing the high speed movies, it was possible to see that at low current, "unmelted" flux sporadically touched the weld pool but at higher current, the flux remained touching the weld pool during the whole time of droplet formation and transfer. It is believed that since the flux has ingredients that contain hydrogen, hydrogen passes through the arc undisturbed, going to the weld bead intact and increasing the hydrogen content in the weld metal. Another important observation is regarding to droplet size. Droplet size increased with increasing current because forces from decomposed gases from the flux could sustain the droplets, retarding their transfer and allowing them to grow.

Keywords: Flux cored arc welding, hydrogen measurement, droplet collection, high speed recording

Introduction

Flux cored arc welding (FCAW) exhibits many important characteristics such as: high productivity, good quality weld and low cost. In the process, the wire is fed continuously into the weld pool, and the shielding is achieved by ingredients inside the tubular wire and from an external gas supply (AWS, 1991). This semi-mechanized process is an excellent candidate for automation.

Despite its many industrial applications, reports (Meyer, 1993; White et al, 1993 and Harwing et al, 1994) have shown that the diffusible hydrogen found in the weld deposits is usually higher than the ones found in the weld metals made using SMAW and GMAW. In practice, the lower values of weld metal hydrogen in GMA and SMA welds are obtained with cleaner wires and drier fluxes and electrode covering. Typical values obtained from different welding process and consumables types are shown in Fig. 1. In this figure, it is possible to observe that even with lower potential hydrogen level, the FCAW process reaches the same or higher weld metal hydrogen level than SMAW. One hypothesis is that this difference could be caused by the flux ingredients inside the wire in the FCAW, which stay unaffected until they reach the arc. In the SMAW process, on the other hand, the baking after electrode manufacturing and the electrode heating during welding reduce significantly the amount of moisture and crystallized water, present in several flux ingredients.

It is well established that hydrogen is the most dangerous gas in steel weld metals because it is the agent for cold cracking. Figure 2 shows examples of the locations where hydrogen cracking can occur in a weld metal. In many cases the fracture can occur catastrophically, 48 hours after welding, and is very difficult to be repaired (Mishler, 1976). It is a consensus that it is important to control the hydrogen content in weld metals made with FCAW process.

The major sources of hydrogen in the weld metal deposited with a tubular wire are (Boniszewski, 1992): moisture in the flux, hydrogenated flux ingredients, residual lubricant on the wire surface, moisture in the shielding gas and in the atmosphere.

This work was developed to evaluate the effect of welding current on hydrogen content in flux cored arc weld metal. The study is based on measurement of diffusible hydrogen using gas chromatography, recording of the electrical parameters - welding voltage and current, collection of metal droplets during welding, and high speed cinematography recording. It is believed that the results obtained in this study will help better understand the electrode melting and metal transfer phenomena and develop ways to control hydrogen content in flux cored arc weld metal by controlling the welding parameters.

Experimental Procedure

A tubular wire, type E71T-1 (AWS A5.20-95) with 1.2 mm diameter, was used in all the experiments. To avoid air contamination when not in use, the wire coil was kept inside a chamber heated by a lamp to temperatures around 130º C. The experimental procedure was divided in two major parts: in the first, experiments were performed to verify the relationship between the welding current and the hydrogen content in the weld metal; In the second, a high speed cinematography recording and droplets collection were made to further evaluate the relationship.

For the first part, standard bead-on-plate welds for hydrogen chromatography analysis according to the International Institute of Welding Recommendation (IIW (a), 1983) were deposited with different welding current. Thirty two experiments, divided in eight groups of four samples were performed. Each group refers to a different welding current. The welds were deposited on ASTM A36 steel coupons with standard dimensions of 30x15x10 ± 0.25 mm and run-on and run-off tabs of 10 x 15 x 45 ± 0.25 mm. Before welding, all coupons were heated in a furnace to temperatures up to 600ºC for 1 hour to remove all residual hydrogen. The coupons were then weighed (W_i – initial weight) and to avoid atmospheric contamination, coupons that were not going to be used immediately were kept in a dessicator.

The initial welding parameters were defined inside a range suggested by the tubular wire manufacturer (ESAB Group Inc., 1996). To select the best parameters to obtain welds with satisfactory characteristics and good aspect, preliminary tests were performed. Table 1 shows the welding parameters adopted after the preliminary tests.

To perform the welding, a semi-mechanized welding system was used. This system maintained constant stick-out and travel speed. For welding, the coupons with the run-on and run-off tabs were fixed in a water-cooled jig, in accordance with the IIW recommendation, to assure uniform heat extraction from the weld region. The welding was performed such that the welds started 10 mm before the coupon, over the run-on tab and finished 10 mm after, over the run-off tab. After arc extinction, the samples were rapidly removed from the jig, immersed and vigorously agitated for 20 to 30 seconds in ice cooled water. The procedure follows by washing the samples in running water, immersing in acetone and air drying before taking them to a hermetically-closed sample holder, purged with argon, for hydrogen collection and analysis. It is important to report that all the post-welding operation did not exceeds two minutes, as suggested in the IIW recommendation. It should also be pointed out that all the operations were performed using gloves and tweezers to avoid samples contamination. For each welding condition, 4 samples were prepared, but only one sample was analyzed at each time. Meanwhile, the other 3 samples were kept inside the hermetically-closed sample holder.

The chromatography equipment (Oerlikon Yanaco – Model H) was calibrated for local atmospheric temperature and pressure following the manufacturer instructions and recommendations. The hydrogen collection was performed in an oven during 72 hours at 45º C, as shown in Fig. 3. After this time, the sample holder was cooled to room temperature and connected to the chromatography equipment. The samples were then weighed (W_f) and hydrogen content, in ml/100g of deposit, was determined using eq. (1).

where:

H_D = Diffusible Hydrogen in weld metal in ml/100 g of deposit,

V = Measured hydrogen volume (ml),

W_f = Final weight(g)

W_i = Initial weight (g)

In the droplets collection experiments, a special device, shown in Fig. 4, was build. The device consisted of a water cooled copper pipe, where the welding was performed with the torch positioned tangent (approximately 45º) to the pipe surface, as shown in Fig. 5. The same semi-mechanized welding system used to prepare the bead-on-plate welds, was used to perform the welding in this part of the research. In these experiments, the arc was established between the wire and the copper pipe surface and as welding proceeds, the droplets touch the pipe surface and fall into a tank filled with water. The tank was divided in nine compartments. The division of the tank into compartments had two objectives: first, to exclude droplets from the beginning and the end of the welding; Second, to compare the number and size of the droplets from the individual compartments (along the welding direction), allowing to specify the error and assuring good quality in the droplets statistical analysis.

With the exception of welding current, the same welding parameters, showed in Table 1, were used in the droplet collection experiments. Welding current varied between 100 and 150 A. For each current, five welding passes were executed. After the set of welding, the droplets were collected, dried and cleaned. Using a pestle and a mortar, the slag was separated from the metal. The droplets were then sieved and classified in six groups of specific diameters. The characteristic volume per group was calculated by arithmetic averaging, using eq. (2) and the characteristic droplet diameter for each welding current was determined using eq. (3).

where

V_c = Characteristic droplet volume;

n_i = number of droplets per group;

V_i = volume of the droplets in a specific group.

where:

d_c = characteristic droplet diameter.

Metal droplet transfer in arc welding can be observed in many ways, however, the most difficult task is to avoid the intense arc light. In this work, arc was positioned between a He-Ne Laser and the high speed camera. Lenses for arc light filtering were positioned in front of the camera so that only the shadow of the wire, droplets and weld pool were visible and recorded.

For the high-speed recording, a high speed ENCORE MAC 8000 S monochrome series motion analysis camera was used. This camera recorded 2000 frames per second. Figure 6 shows the apparatus used for high speed recording. A small table, driven by a DC motor, was used to move the plate under the torch, keeping the arc steady in front of the camera while the welding was performed. The welding parameters used in this part of the experimental procedure were the same used before, showed in table 1.

After welding, the images were captured, converted to digital pictures, frame by frame, and transferred to a computer. Figure 7 shows the assembly for image captures and manipulation.

Results and Discussion

For the correlation with diffusible hydrogen and to assure the repeatability in the welding parameter, arc voltage and welding current were monitored during the experiments. The monitoring was performed during 5 intervals of 4 seconds between intervals of 3 seconds, allowing the collection of up to 10,000 points of each parameter in the form of oscilograms. In general it was possible to observe that, even though it was tried to keep the voltage constant, small fluctuations were noticed. For each group of experiment the average values of voltage and current for the 10,000 values monitored for the four samples of each group were adopted. Figure 8 shows the relation between the average arc voltage and the average welding current for the eight groups of experiments.

It is important to notice that the errors obtained between the four samples, from each group of experiment, were insignificant. From Fig. 8, it is possible to observe that the linear regression of the points gives a slope value of -0.02 volts/amp, in accordance with the nominal values expected for the type of power source used (Cary, 1989). In other words, the average voltage changes were within the minimum range achievable by the equipment and it is acceptable to say that the voltage stayed almost constant during the experiments.

Table 2 shows the results of diffusible hydrogen in the weld metal obtained for the eight settings of current. The results corresponded to the average values for each parameter or between the four samples welded with each condition. The values are graphically represented in Fig. 9. Linear, quadratic and cubic regressions were performed to verify the trend. The four curves represented regressions of minimum quadratic type, for the correspondent polynomial degree, minimizing the function:

where:

D = quadratic mean deviation

y_i = ith ordinate obtained experimentally

f(x_i) = value of the interpolant function for the ith abscissa

N = Number of points obtained experimentally

Table 3 shows the equations for the constant, linear, quadratic and cubic approximations, with the respective deviations, calculated according to equation 2. From this table it is possible to observe that the deviations for the linear, quadratic and cubic approximations are very close to each other. In other words, the quadratic and the cubic approximations add very little to the linear approximation. The results seem to indicate that the phenomena of diffusible hydrogen increase with welding current is a true linear relationship with proportionality constant of approximately 0.7 or 70%. This finding is important for one to predict the increase of diffusible hydrogen in weld metal made with FCAW if the variation of welding current is known. It can also be used in projects to determine the safety factor for welded structures.

However, one question remains unanswered: why did hydrogen content increase since the measured arc voltage stayed almost constant during the experiments? One should then expect that the hydrogen increase effect should be related only to welding current. For solid wire (GMAW), for example, it is well known that as current increases the gas (oxygen, nitrogen and hydrogen) contents in weld metals tend to decrease (Lancaster, 1993). This observation is usually associated with the droplet size that diminishes as current increases (Liu et al, 1989; Brandi et al, 1991, Wang et al, 1995 and Pistorius and Liu, 1997). At high current, with constant arc voltage, the droplet volume is so small that they can not carry much gas inside of it. Add to it the fact that droplet acceleration from the wire tip to the weld pool is higher at high current and there is not enough time for the droplets to catch gas atoms during the flight through the arc and carry them, hydrogen for example, to the weld pool.

Then, from the discussion above, for FCAW it should be expected a significant increase in droplet size when current increases and this increase should be of at least 70%, to explain hydrogen increase with current. To evaluate this hypothesis, droplet collection was performed. Figure 10 shows samples of droplets collected at the currents of 105, 120, 135 and 150 A, immediately after they were removed from the collector tank. Figure 11 shows the droplets classification in six groups of specific diameter.

A first look at Fig. 10 gives the impression that at current of 150 A, the droplets are far bigger than the ones at 105 A. However, Table 4 shows the results of characteristic droplet diameter calculated for each current (using equation 3) after arithmetic averaging the characteristic volume per group (using equation 2), also showed in Fig. 11. In this table the relation between the characteristic diameter (d_c) and the wire diameter (d_w) is also presented for each current. Figure 12 shows the results graphically.

Observing Fig. 12, one can say therefore that the increase in droplet size is relatively insignificant regarding current increases. While current increases approximately 50% (from 105 to 150 A), droplet size increases only 20%. In other words, the increase in the droplets size can not explain the 70% of hydrogen increase with current. Therefore, another phenomena must be the responsible for the hydrogen increase. One important point regarding Fig. 12 is the ratio between droplet diameter and wire diameter. The ration d_c/d_w is always bigger than 1, indicating that the transfer mode for this range of current is of globular type, according to the IIW definition ( IIW (b), 1976).

To further evaluate the effect, the high speed movies were made. Figure 13 shows sequences of frames for currents of 105 A, 150 A and 200 A. After analysis, it was clear the cause of hydrogen increase with current: at high currents, the cored flux touches the weld bead. At 150 A the contact is sporadic but at 200 A, it stays touching during the whole time of droplet formation and transfer. Since the flux has ingredients that contain hydrogen, hydrogen passes through the arc without interacting with the arc, going into weld bead intact and increasing the hydrogen content in the weld bead. Figure 14 and 15 show more detailed frames where it is possible to see the flux column touching the weld bead, breaking down just before transferring. A generous amount of the flux is transferred to the weld pool as well.

Another important observation from the recorded movies is regarding the droplet size. At low current the metal sheath melts and a portion of the flux, such as a cone, projects to the weld bead direction while the droplets form laterally to the wire and fall to the weld pool as a certain size is reached. As current increases, the wire speed increases and the flux pass from cone to a cylinder and further project to the weld pool, as described above. In this condition, a larger portion of the flux is now in contact with the arc. It is believed that buoyancy forces from gases decomposing in the flux can sustain the droplet, retarding its transfer and allowing it to further grow. It is possible to observe that even with higher melting rate, the size of the droplets are bigger at higher current than at lower current, confirming the results obtained at the droplet collection. Figure 16 shows a schematic model of this phenomena observed.

Conclusions

According to the results of this investigation, the major conclusions are presented as follows:

Acknowledgments

The authors acknowledge the financial support of the FINEP, Financiadora de Estudos e Projetos – under contract RECOPE 779831700, reference 1567/96. They also acknowledge the support from CNPq Conselho Nacional de Pesquisa – Brazilian Government.

Article received November, 2001

Technical Editor: Alisson Rocha Machado

Publication Dates

History