Abstract
Optimization of the weld quality and productivity requires in-process identification and simultaneous regulation of several thermal characteristics of the joint. Since in traditional single-torch welding only a few process variables can be modulated in real-time, multiple source configurations are implemented by a rapidly reciprocated (timeshared) GTAW torch to obtain decoupled control of the weld geometry, structure and properties. Further, to widen the range of achievable weld features, a scanning motion of the torch on the entire part surface generates the necessary heat distribution for any specified thermal field in the weld, which is observed through surface temperature measurements. Analytical, numerical, and experimental thermal modeling techniques are employed for the design of multivariable adaptive and distributed-parameter controllers, applied to girth and flange welding simulations, and tested in seam pipe welding experiments, for rejection of process disturbances and for weld quality regulation performance.