6 TH INTERNATIONAL CONFERENCE ON TRENDS IN WELDING RESEARCH, APRIL, 2002 1 Is there evidence of determinism in droplet detachment within the gas metal arc welding process? Charles R. Tolle, Randall A. Laviolette, Herschel B. Smartt, Kevin L. Kenney, David P. Pace, John W. James, Arthur D. Watkins, Thomas Wood, and Daphne L. Stoner Idaho National Engineering and Enviromental Laboratory P.O. Box 1625 Idaho Falls, ID 83415-2210 USA Abstract— For many years, the welding community has assumed that the detachment events within Gas Metal Arc Welding (GMAW) process are stochastically driven. With this in mind, we set out to better understand these mechanisms and what might drive the detachment events. Through the investigation of 6061-Aluminum GMAW, a number of new insights into globular, and spray transfer modes have been found. By using windowed Fourier transform analysis, akin to Wavelet Analysis, we present evidence that the GMAW process undergoes a shift from the globular mode fun- damental frequency to its secondary harmonic frequency as the GMAW process shifts from globular to spray mode transfer. This breakdown and shift from primary to secondary fundamental frequencies and its reverse are shown to occur often within the transition-welding mode in aluminum GMAW. Furthermore, there is partial evidence that this shift of fundamen- tal frequency modes from secondary to the third fundamental frequency continues with the transition from the spray to streaming transfer. The ev- idence for determinism as the driving force behind GMAW detachments is strengthened through chaotic systems analysis, e.g. surrogate testing using ApEn statistics. Keywords—GMAW; Windowed Fourier Analysis; ApEn; Determinism I. I NTRODUCTION In gas metal arc welding (GMAW), the use of a consum- able electrode results in a degree of coupling between heat and mass transfer. Approximately 80 to 85% of the electrical en- ergy consumed by the process is transferred to the weldment as heat [1]; approximately one-half of this transferred energy is transported as sensible and superheat molten drops of metal [1]. Consequently, the manner in which the molten drops are trans- ferred to the weldment are of interest. An IIW classification of metal transfer listed six different transfer modes for GMAW [2]. More recent convention lists three modes, in the order of in- creasing current: short circuiting transfer, globular transfer, and spray transfer [12]. This list neglects an additional mode, well recognized by researchers, called streaming transfer or spray- streaming transfer which occurs at higher current levels. It is also known that rotating transfer can occur at even higher cur- rent levels [3], but it is seldom seen in normal welding practice. As for this work, we focus on the globular and spray transfer modes. This work was supported by the U.S. Department of Energy, Assistant Sec- retary for Environmental Management; U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences; and U.S. Department of Energy, Of- fice of Industrial Technology under DOE Idaho Operations Office Contract DE- AC07-99ID13727. Charles R. Tolle is the corresponding author (Email:tollcr@inel.gov; Phone 208-526-1895). It has long been recognized that the mode of droplet trans- fer in gas metal arc welding plays an important role in heat and mass transfer of the process. The various modes of transfer have been identified and mapped in parameter space and static mod- els of detachment have been formulated. However, it is only recently that attempts have been made to understand the physi- cal basis for the dynamics of droplet transfer. One of the chal- lenging questions in understanding droplet transfer dynamics is determining if droplet transfer is a deterministic or stochastic process. In this work, we look at the frequency content of the electri- cal signals in GMAW to see what information is present about the metal transfer mode. We have chosen to focus on analyzing the voltage measurements between the contact tip to the work piece. In the sections that follow, we outline the experimental setup; show a correlation between Fourier analysis peaks and droplet detachment times; present a comparison between ex- perimental data and a deterministic GMAW model [4],[5]; and present some initial chaotic analysis used to determine if a signal might be from a deterministic process. The process we under- take within this paper provides a guide for investigating deter- minism within other dynamic processes of interest. II. ALUMINIUM GMAW EXPERIMENTS Within the course of several different welding projects, we became interested in sensing the droplet detachment events through simple electrical measurements for possible inclusion within an advanced GMAW control scheme. As we investigated different welding materials, e.g. steel and aluminum, we found that the electrical signals measured within the aluminum process appeared qualitatively cleaner than those for steel. With this knowledge in hand, we decided to focus on gas metal arc weld- ing of 6061 aluminum bead on plate with 4043 filler wire using an industrial grade (100%) Argon gas shield. We tested several basic welding setups with similar results. For the experimen- tal data presented here, we choose the 4043 electrode diameter to be 0.76mm, the welding travel speed as 10 mm s , the constant voltage setting as 20 volts, the contact tip to work piece distance as 20mm, the 6061 base plate as 1 4 inch sheet, a gas flow rate of 30scf h, a weld time of 5 seconds, a data acquisition sample rate of 5000Hz, and we allowed the wire feed speed to range