Inverse Thermal Analysis of Welds Using Multiple Constraints and Relaxed Parameter Optimization

Abstract

Aspects of a methodology for inverse thermal analysis of welds are examined that provide for relaxed model-parameter optimization. These aspects are associated with the inherent insensitivity of temperature fields, obtained by inverse analysis, to local shape variations of constrained boundaries within these fields. The inverse analysis methodology is in terms of numerical-analytical basis functions for construction parametric temperature histories, which can be adopted as input data to computational procedures for further analysis. In addition, these parametric temperature histories can be used for inverse thermal analysis of welds corresponding to other process parameters or welding processes whose process conditions are within similar regimes. The inverse thermal analysis procedure provides for the inclusion of volumetric constraint conditions whose two-dimensional projections are mappings onto transverse cross sections of experimentally measured boundary conditions, such as solidification and transformation boundaries, and isothermal surfaces associated with thermocouple measurements. Issues concerning relaxed parameter optimization are discussed with respect to inverse thermal analysis of Ti-6Al-4V pulsed-mode laser welds using multiple constraint conditions.