Methodology of determining material parameters based on optimization techniques (Q2043174)

The goal of numerical modelling while producing new devices is to avoid creating expensive prototypes in order to cheapen the production. The problem is that the characteristics of the materials involved in the modelling may be unknown in the temperature range of interest and their experimental determination is often impossible. For example, the heat conduction and the specific heat of some metal or alloy may be unknown above its melting point. To find such temperature-dependent material properties the authors suggest the following method. The temperature dependencies of the desired material characteristics are calibrated, i.e., compared with analogous properties of similar material. Then the desired function of temperature begins to depend on several adjustable parameters which must be found by solving an optimization problem. In turn, during optimization some process is modeled which involves the materials under study. The temperature field at this process is calculated in some moments of time and compared with experimental data. The adjustable parameters are selected to minimize some quality functional describing the difference between the experimental and calculated temperatures. The method described is illustrated by two processes: laser welding and laser cladding. Several numerical optimization methods are discussed and compared.