Thermal Manufacturing Process Control by Lumped Mimo and Distributed-Parameter Methods

Abstract

A variety of geometric, material structure, and stress/distortion attributes are needed to characterize the quality of thermally manufactured products. Because of in-process sensing difficulties and transportation lags, these features must be regulated in real time through appropriate thermal outputs, measured by non-contact infrared pyrometry. In thermal processes with a localized, sequentially moving heat source, the necessary heat input distribution on the part surface is supplied by an innovative timeshared or scanned torch modulation, in a raster or vector pattern. A unified lumped multivariable and a distributed-parameter quasilinear modeling formulation provide a design methodology and real-time reference for the development of finite- or infinite-state adaptive thermal control systems. These controllers modulate the power and motion of a single torch, supplying distinct concentrated heat inputs or a continuous power distribution on the part surface, so as to obtain the specified thermal characteristics or the entire temperature field. These regulation strategies are computationally tested and implemented experimentally in arc welding, but their applicability can be extended to a variety of thermal manufacturing processes.