An Effective Inverse Heat Transfer Procedure Based on Evolutionary Algorithms to Determine Cooling Conditions of a Steel Continuous Casting Machine

Abstract

The inverse modeling of heat transfer is a useful tool in analyzing contact heat transfer at the ingot surfaces during the continuous casting process. The determination of the boundary conditions involves an experimental work consisting in the evaluation of the thermal history, generally at the casting surface, experimentally provided by infrared pyrometers. Additionally, numerical simulations, based on the solution of the 2D transient heat conduction equation, are performed in order to be inversely solved in response to the measured thermal data furnished by the sensor. Due to computational time consumption during simulations in searching cooling conditions, this work proposes an interaction between natural inspired algorithms, called evolutionary algorithms, and the numerical model in order to speed up the searching process. The present work aims to compare three algorithms, namely genetic algorithm, improved stochastic ranking evolutionary strategy, and evolutionary strategy with Cauchy distribution. The latter develops a metaheuristic version of an evolutionary strategy workflow, using a Cauchy random number function to generate each individual, instead of the usual uniform distribution function available in almost all programming languages. The surface temperature, solid shell, and molten pool profiles from the determined cooling conditions are analyzed in terms of casting quality.