Professional Documents
Culture Documents
Pulse Mig Welding
Pulse Mig Welding
RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT
ABSTRACT. An efficient method of iden- typically more sensitive to parameter determine optimum power supply set-
tifying power supply pulsing parameters changes. As a result, careful parameter tings for various applications (Ref. 3). No
for pulsed gas metal arc welding based selection is important. To address this, efficient method has been described in
on statistical experimental design is pre- “synergic” pulsed GMAW power sup- the open scientific literature for estab-
sented. Fractional factorial screening ex- plies are commonly used in industry lishing viable pulsing parameters in
periments are combined with D-optimal (Refs. 1, 2). In these systems, the rela- GMAW-P. Efficient use of statistical de-
experimental designs to allow the user to tionship between a digitally controlled sign of experiments (DOE) techniques,
develop an accurate wire feed rate model wire feed rate and the pulsing parameters such as fractional factorial design and D-
for varying pulsing conditions and to is managed by microprocessor-based optimal design, allows development of
characterize the desirable one droplet look-up tables. While easy to use, these an empirical methodology, incorporating
per pulse (ODPP) operating region for a systems do not give the user much flexi- a scientific approach to welding proce-
given wire type and diameter. Equations bility. Thus, they are limited in their ca- dure development. The D-optimal de-
defining the wire feed rates and time at a pabilities for special applications such as sign technique, with its flexibility in
given peak current required for ODPP very thin sections or very high speed. defining the experimental space, is use-
transfer are presented. Compared to con- In addition to operating in the syner- ful for modeling the nonlinear aspects of
ventional techniques, a very small num- gic mode, in which the user typically se- arc welding processes. This is not easy to
ber of experiments is required. A power- lects only the material (i.e., steel, alu- do with conventional experimental de-
supply-dependent approach and a more minum, stainless steel) to be welded and sign techniques where requirements for
generic method employing measured a power level, many modern power sup- orthogonality may make it more difficult
rather than nominal current values are plies also allow users to select pulsing pa- to characterize the process. The purpose
presented. Joints produced using this ap- rameters. However, the task of determin- of this paper is to present an experimen-
proach are evaluated and found to meet ing what the welding parameters should tal methodology for identifying and opti-
applicable bead geometry standards. be can be very time consuming and ex- mizing certain characteristics of weld
perimentally intensive, as evidenced by pulsing parameters.
Introduction product literature for some synergic
power supplies, which notes that mil- Background
Identifying suitable combinations of lions of experiments were conducted to
welding parameters for use with pulsed Salter and Doherty (Ref. 4) identified
gas metal arc welding (GMAW-P) can be the operating boundary for submerged
a time-consuming process, involving arc welding and GTA welding experi-
considerable trial and error. The primary mentally using design of experiments.
parameters to be identified are wire feed They developed empirical relations for
KEY WORDS
rate, peak current, background current, bead geometry parameters and sug-
duty cycle and pulsing frequency. As il- gested that empirical equations can be
Fractional Factorial
lustrated in Fig. 1, the welding parame- used to pick robust solutions or welding
D-Optimal
ters are more numerous than in conven- procedures, so that the visual quality ac-
Pulsing
tional GMAW, and the process is ceptance criteria are satisfied when the
GMAW-P weld process inputs vary between given
Droplet Transfer levels.
S. SUBRAMANIAM and D. R.WHITE are with Peak Current Several other researchers have used
Ford Research Laboratory, Dearborn, Mich. J. Background Current factorial techniques to identify the effects
E. JONES is with N.A. Technologies, Golden, Gas Metal Arc
Colo. D. W. LYONS is with Department of Me- of main factors and their interactions in
chanical and Aerospace Engineering, West controlling weld quality. The experimen-
Virginia University, Morgantown, W.Va. tal results are used to develop regression
Fig. 4 — Wire feed rate as a function of power input for sev- Fig. 5 — Relation between peak and background conditions
eral sets of pulsing parameters that produce ODPP transfer. for ODPP using nominal values of Ip and Ib shown in Table 4.
Numbers on the points correspond to those on Table 4.
Fig. 8 — Weld bead of 1.5-mm, 1.5-mm lap joint. Fig. 9 — Weld bead of 2-mm, 3-mm lap joint.
determined experimentally. For alu- W = 0.14*(Ip) + 1.08*(Ib) + the desired one droplet per pulse (ODPP)
minum resistance heating is assumed to 4.18*(D) + 0.03*(F) + 0.8*(Ltp) (5) transfer over a wide range. To minimize
be negligible. the experimentation needed, D-optimal
This equation has been modified for where Ip = peak current, Ib = background designs are used, as discussed in the next
GMAW-P by integrating over one pulse current, D = duty cycle, F = pulsing fre- section.
cycle as (Ref. 28) quency, Ltp = contact tip to plate dis-
tance. Metal Transfer Mode Experiments
W(t) = Instantaneous melting rate Statistical analysis of variance was
carried out for these results. The R2 value D-Optimal Designs
W = ∫W(t)dt was found to be 0.9992, which shows
cycle that the model fit the experimental data The most commonly used experiment
accurately. The equation shows that the designs are cubic in nature and attempt
≈ (WpTp + WbTb)F (4) background current and duty cycle are to generate orthogonal data. However, in
the most significant factors affecting wire some applications, irregular experimen-
where Wp = wire melting rates in peak feed rate, in the range of parameters stud- tal regions are found due to physical or
current, Wb = wire melting rate in back- ied in the two-level experiment. Though economic constraints. This problem is
ground current, F = pulse frequency, a higher order quadratic equation would often encountered in attempting to con-
Tp=peak time, Tb = background time. show the effect of interactions, it would duct designed experiments on welding
This equation is not sufficiently accu- be more complicated. The linear fit to the processes. It is often impossible to estab-
rate to be used to generate pulsing para- experimental data presented here, how- lish an arc, or melt-through may occur
meters on an a priori basis for GMAW-P ever, provides a useful tool to quickly es- under certain experimental conditions
of aluminum over a wide range. The pri- tablish a wire feed rate for a stable weld- needed to satisfy the typical orthogonal
mary reason for this inaccuracy is a fail- ing operation. experimental design. In these circum-
ure to account for the effect of pulsing In some parameter ranges, the wire stances, the experimenter is forced to ei-
frequency and electrode extension in feed rate predicted by Equation 4 and the ther limit the range of the experiment,
aluminum, but the model also does not model presented here can be quite dif- making it difficult to accurately charac-
account for power supply dynamics. The ferent, as shown in Fig. 2. Figure 2 shows terize the entire process space, or to re-
actual current values can be quite differ- the predicted wire feed rates under the sort to methods of describing “bad data,”
ent from the nominal values in the peak same conditions using Equation 4 and which cause inaccuracies in the regres-
phase. As a result, wire melting during Equation 5 at different pulsing frequen- sion anaysis (Refs. 25, 29).
peak and background cannot be simply cies. This difference could result in an As a result, it is sometimes useful to
added assuming a square wave pulse. unstable operation. employ other types of experiment de-
The values of wire feed rate obtained Once the wire feed rate required for signs to model a process. D-optimal de-
in the two-level screening DOE de- stable operation as a function of the puls- signs offer the flexibility to solve such
scribed earlier were used to develop a ing parameters is determined from the re- problems by allowing the experimenter
linear wire feed rate model. The follow- gression analysis, the next step is to iden- to modify the design space by including
ing equation was found to fit the data: tify the pulsing parameters that produce and excluding operating points, an im-
( )
496.1 * 1 − e – 0.003* I b Tb +
Tp = 270.1 /I (7)
p
( b b − ) +1
2
I T 182 .2
8423.5
This can be used to define the minimum
time at peak required for droplet detach-
ment at a desired peak current level.
Table 4 shows the pulsing parameters
that result in ODPP. Table 4 also shows
possibility in classical designs. Exclusion at three levels were required to roughly the average currents over a fairly wide
of individual runs from the experimental characterize metal transfer behavior range of 108–250 A and the correspond-
design is generally accomplished by using this approach, as shown in Table 3 ing wire feed rates. Depending on the
specifying an exclusion formula to re- (Ref. 23). heat input and deposition rates required,
move the infeasible points from the orig- a user can start with a particular proce-
inal full factorial candidate set. Inclu- Mapping of One Droplet Per Pulse (ODPP) dure and use Equation 7 as a tool to find
sions are usually experimental settings
parameters that provide ODPP at a par-
for which data already exist prior to the As noted previously, it is generally re- ticular average current.
experiment, although inclusion formulas ported that it is important in GMAW-P to Like the power law approaches used
can also be specified, leading to greater establish pulsing parameters that provide by others (Refs. 18, 19), this equation
experimental efficiency (Refs. 25, 29). ODPP transfer (Refs. 14–19). The ODPP
The D-optimal approach also supports does not incorporate frequency — the
condition is also reported to result in number of droplets transferred in a single
smaller three or more level experiments. In welds with minimal defects and spatter
a conventional factorial design of experi- pulse, strictly a function of the relation-
(Ref. 30). Empirical relations of the form ship between time at peak current and
ments, the number of experiments neces- Ip Tp = C have been proposed by previ-
sary for orthogonal design can become peak current values (assuming Ibtb
ous researchers for work done at low <<Iptp). Frequency of pulsing will affect
prohibitive as the number of levels in-
background currents in attempting to deposition rate by controlling the num-
creases. For nonlinear processes such as
characterize the ODPP transfer regime. ber of times in a given time interval in
welding, this ability to more efficiently
By analyzing the high-speed films and which these droplets transfer. This Equa-
model three or more levels is another im-
voltage signals, experiments that resulted tion 7 provides a simulation tool that can
portant advantage of D-optimal designs.
in one drop per pulse were identified for be used to study different pulse parame-
Though calculated effects from D-optimal
experimental designs are not perfectly or- developing an empirical model. Unlike ters before starting any experimental
thogonal/independent, errors are mini- previous efforts, this model takes into ac- work. It decreases trial and error experi-
mized, and this is considered an accept- count the background conditions (Ref. mentation required to develop welding
able trade-off for the flexibility and 23). As the actual values of current and procedures and provides a systematic
efficiency gained using D-optimal designs. voltage are used, it is a generic model for methodology to identify suitable pulse
In our work, D-optimal experiments this wire diameter and composition. parameters. The peak current and back-
were used after the wire feed rate model ground conditions can be fixed, and the
was established to determine the condi- Results of Metal Transfer Experiments wide range of required peak time can be
tions for one droplet per pulse transfer. computed from the equation. Once the
The parameter levels of the four pulsing To be consistent with previous work peak time is known, the pulsing fre-
parameters used in this experiment are (Refs. 18, 19), the relationship between quency to obtain ODPP conditions can
shown in Table 2. Only 19 experiments the peak and background conditions for be calculated.
RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT/RESEARCH/DEVELOPMENT
be controlled for maximizing productiv-
able for a given application. Depending Welding Parameter Selection ity. D-optimal design allows the ODPP
on the application, minimizing heat region to be characterized using very few
input, maximizing deposition rates or in- The model given by Equation 6 is a experiments.
creasing travel speed may be the major generic model applicable to any power This methodology can help minimize
concern. supply because it is based on the actual the power supply dependence of the re-
In this work, minimizing heat input to heat input delivered to the weld. For a sults by considering actual pulsing con-
the thin section material and obtaining specific power supply, it is easier to de- ditions. The data needed can be obtained
high productivity were both issues to be velop welding procedures with the using only an oscilloscope to collect ac-
addressed. Figure 4 shows the relation- model developed from the programmed tual current values as a function of time.
ship between the power input and wire values, given by Equation 8. The results However, nominal values can also be
melting rate during ODPP transfer for this of the methodology presented here were used. The one droplet per pulse condi-
wire. To minimize the heat input, the used to develop welding schedules. tion can be characterized by detecting
pulse parameters can be set such that These conditions lie close to the curve on droplet detachment in the voltage sig-
power input to the process is minimized Fig. 5. If current values are recorded nals. This technique also provides an ef-
for a particular deposition rate. Figure 4 while welding is carried out, the actual ficient way of conducting experiments
shows how the melting rate increases values of IpTp and IbTb calculated will lie for process development, allowing any
with power input, which is again a func- on the curve in Fig. 3. Figure 6 shows the user to rapidly determine viable pulsing
tion of the pulsing parameters. All of the bead geometry of a weld in a lap joint on schedules.
data points represent ODPP transfer con- a 1.5-mm extrusion with ODPP condi-
ditions using different combinations of tions, which met applicable quality re- Acknowledgments
pulsing parameters. Between points 2 quirements for the products of interest in
and 3 (which have different pulsing pa- this study (‘S’ is the travel speed) (Ref. The contributions of William Weber,
rameters but the same droplet transfer 31). Figure 7 shows welding of an alu- David Scholl, Richard Allor, Rick Baer
frequency) the change in power input is minum extrusion to a casting with a pro- and Joe Williams to the experimental
only 1.4%, but it results in an increase in cedure resulting in ODPP, with the thin- work described here are gratefully ac-
deposition rate of about 11%. Looking at ner section at the bottom. knowledged.
points 3 and 4 on the plot, an increase in Although ODPP is usually considered
total power input of about 16% results in to be the ideal in GMAW-P (Ref. 32), in References
an increase in melting rate of only 3%. a number of cases, good quality welds
These results are directly related to the were made under conditions of more 1. Amin, M. 1981. Synergic pulse MIG
metal transfer behavior, and it is impor- than one droplet per pulse. Figure 8 welding. Metal Construction 13(6): 349–353.
tant to understand the influence of all the shows a weld joining a 1.5-mm extru- 2. Rogers P. 1985. Synergic MIG in per-
pulsing parameters when optimizing the sion. This welding condition resulted in spective. Metal Construction 17(9): 602–604.
process. By allowing one to characterize two drops per pulse. The IpTp value for 3. Nacey, T. J. 1993. Fourth generation in-
the range of parameters within which we this condition lies above the curve shown verters add artificial intelligence to the control
can obtain ODPP with minimal trial and in Fig. 5. This shows that the peak con- of GMA welding. Welding Journal 72(1):
error experimentation, this work pro- ditions in this case were higher than nec- 31–34.
vides an efficient starting point for pro- essary for ODPP, but still produced a 4. Salter, G. R., and Doherty, J. 1981. Pro-
cedure development. good quality weld. Figure 9 shows the cedure selection for arc welding. Metal Con-
weld bead for joining 2–3-mm extru- struction 13(9): 544–550.
Modeling Using Programmed Values sions. The peak conditions in this case 5. Harris, P., and Smith, B. L. 1983. Facto-
are also higher than needed for ODPP. rial techniques for weld quality prediction.
The data presented here are based on Metal Construction 15(11): 661–666.
continuously measured values of Ip and Conclusions 6. Pandey, S., and Parmer, R. S. 1990.
Ib at the weld. Actual, rather than nomi- Mathematical models for predicting bead
nal values, are not always available, but A method has been developed that geometry and shape relationships for MIG
this methodology could be applied using provides an efficient approach for select- welding of aluminum alloy 5083. Proceedings
the programmed pulse parameters for a ing pulsing parameters while requiring of the 2nd International Conference on Trends
particular power supply, as shown here considerably less experimentation than in Welding Research, eds. S. A. David and J.
for a Miller Maxtron 450. A plot of the re- conventional approaches. This is a five- M. Vitek, pp. 37–41, ASM International, Ma-
lationships between IpTp and IbTb for step process consisting of screening ex- terials Park, Ohio.
ODPP conditions using programmed periments, development of wire feed rate 7. McGlone, J. C. 1982. Weld bead geom-
values is shown in Fig. 5. model, D-optimal experimental design, etry prediction — a review. Metal Construc-
A simple quadratic polynomial is a identification of ODPP region from metal tion 14(7): 378–384.
good approximation to the experimental transfer analysis and optimization for a 8. Thorn, K., Feenstra, M., Young, J. C.,
data, as given by Equation 8. particular application. Lawson, W. H. S., and Kerr, H. W. 1982. The
Combining a two-level DOE with the interaction of process variables — their influ-
IpTp = -0.0044(IbTb)2 + D-optimal experimental design provides ence on weld dimensions in GMA welds on
2.97*(IbTb) - 60.13 (8) a way to identify the operating envelope steel plate. Metal Construction
while incorporating the physical con- 14(3):128–133.
A comparison of Figs. 3 and 5 illus- straints of the process to improve experi- 9. Subramaniam, S., Misra, S., Pandey, S.,
trates why welding procedures are so mental efficiency. and Sharma, P. B. 1992. Optimization of weld-
ASM International, Materials Park, Ohio. steady and pulsed current GMA welding Al- 353-s.
10. Smati, Z. 1986. Automatic pulsed MIG Mg alloys. Proc. Sixth Int. Conf. on Aluminum 28. Quintino, L., and Allum, C. J. 1984.
welding. Metal Construction 18(1): 38R–44R. Weldments, pp. 207–214. American Welding Pulsed GMAW: interations between process
11. Amin, M. 1983. Pulse current parame- Society, Miami, Fla. parameters — part 1. Welding and Metal Fab-
ters for arc stability and controlled metal trans- 20. Reference Manual. FORD ACCELL rication 3: 85–89.
fer in arc welding. Metal Construction 15(5): Welding System, AMET, ID. 29. Snee, R. D. 1985. Computer aided de-
272–277. 21. Allemand, C. D., Schoeder, R., Ries, D. sign of experiments — some practical experi-
12. Needham, J. C. 1965. Pulse controlled E., and Eagar, T. W. 1985. A method of filming ences. Journal of Quality Technology 17(4):
consumable electrode welding arcs — general metal transfer in welding arcs. Welding Jour- 222–236.
principles and operating characteristics. nal 64(1): 45–47. 30. Takeuchi, Y., and Shinoda, T. 1991.
British Welding Journal 4:191–197. 22. Subramaniam, S., White D. R., Jones, Spatter and blowhole formation in pulsed gas
13. Boughton, P., and Lucey, B. A. 1965. J. E., and Lyons, D. W. 1998. Analysis of arc shielded metal arc welding. Materials Science
The use of pulsed current to control metal voltage, current, and light signals in pulsed gas and Technology 7(9):869–876.
transfer in welding. British Welding Journal 4: metal arc welding of aluminum. Science and 31. Subramaniam, S. 1996. Process mod-
159–166. Technology of Welding and Joining eling and analysis for pulsed gas metal arc
14. Matsuda, F., Ushio, M., and Tanaka, Y. 3(6):304–311. welding of an aluminum automotive space-
1983. Metal transfer characteristics in pulsed 23. Subramaniam, S., White, D .R., Jones, frame. Ph.D. dissertation. Morgantown,
GMA welding. Transactions of JWRI 12(1). J. E., and Lyons, D. W. 1998. Droplet transfer W.Va., West Virginia University.
15. Matsuda, F., Ushio, M., Tanaka, Y., in pulsed gas metal arc welding of aluminum. 32. Needham, J. C., and Carter, A. W.
Itonaga, K., and Yokoo, T. 1984. Pulsed Welding Journal 77(11): 458-s to 464-s. 1965. Material transfer characteristics with
GMAW: one-drop-transfer and process para- 24. Statgraphics PLUS. Ver. 7, Manugistics. pulsed current. British Welding Journal 5:
meter. Transactions of JWRI 13(2): 15–20. 1993. 229–241.
16. Ueguri, S., Hara, K., and Komura, H. 25. RS/Discover. Statistical Appendices.
1985. Study of metal transfer in pulsed GMA BBN Software Products Corporation. 3-1:26,
welding. Welding Journal 64(8): 242-s to 250-s. 1989.
17. Kim, Y. S., and Eagar, T. 1993. Metal 26. Subramaniam, S. 1996. Wire feed rate
transfer in pulsed current gas metal arc weld- model for pulsed gas metal arc welding of alu-
ing. Welding Journal 72(7): 279-s to 287-s. minum. Winner of Henry Granjon student
18. Allum, C. J. 1983. MIG welding — paper competition. International Institute of
time for a reassessment. Metal Construction. Welding.
Full-color reprints of the subject paper, originally published in the February 1999 Welding Jour-
nal, are now available at no charge from the AWS Publication Services Division. This reprint
includes a full set of color illustrations, as originally submitted by the authors. For a copy, please
call the Welding Journal offices at (800) 443-9353, Ext. 348, or fax us at (305) 443-7404.