Process modeling and analysis for pulsed gas metal arc welding of an aluminum automotive spaceframe.

Procedure development for welding of thin section (1-4mm) aluminum, using Pulsed Gas Metal Arc Welding (PGMAW) requires considerable understanding of the process physics. Selection of ideal pulse parameters for different joint designs is a non-trivial task. Several aspects of the process were studied, in order to have a better fundamental understanding of the process. Welding was conducted using 1.2 mm diameter 4047 aluminum electrode and argon shielding gas. An extensive collection of high speed pictures were taken over a wide range of pulse parameters using a laser shadowgraph setup to study the metal transfer behavior. Current, voltage and arc light emission signals were concurrently recorded. Process modeling of lap joints using D-optimal experimental design and neural networks is performed. This method allows to model processes which are difficult or not possible to model using conventional statistical methods, over a wide range of heat input conditions. Analysis of current, voltage and light signals correlated with the high speed films, shows that the power supply dynamics has a significant influence on the process. Droplet detachments were detected from the voltage signals. Limited experiments on control applications showed that voltage sensing is a promising technique for joint tracking in welding of thin section aluminum. The metal transfer behavior was analyzed from the high speed images. A new model is developed, to identify the one drop per pulse condition in aluminum, which also considers the background conditions. Computations of energy input per pulse and static forces is performed. The energy input explains the droplet transfer mechanism under different conditions. Appreciable droplet acceleration observed in aluminum is analyzed. Wire feed rate was found to be a sensitive parameter to establish stable operation in PGMAW. Conventional models based on energy balance and empirical relations do not yield good results when applied over a wide range of pulse parameters. New linear and quadratic models are developed which accurately predict the wire feed rate. Based on the results of this work an engineering methodology, using efficient experimentation and process physics is proposed for process development using PGMAW