Evolution of charging programs for achieving required gas temperature profile in a blast furnace

This paper presents an approach by which charging programs in the blast furnace can be evolved. The core of the method is a mathematical model, which on the basis of a given charging program estimates the two-dimensional distribution of burden layers in the shaft. A gas flow model uses this information to estimate the gas distribution, applying a simplified treatment of the conditions in the upper shaft. The aim is to find the charging program that gives a state of the furnace shaft matching a target for the radial temperature profile at the level of an in-burden probe. This is accomplished by applying a genetic algorithm that makes an efficient search among the huge number of potential charging programs, executing the burden and gas flow models in the function evaluations. The method is illustrated by six cases, where targets for the gas temperature distribution are given and the genetic algorithm evolves the charging sequence and the chute settings for the dumps. It is demonstrated that the algorithm efficiently can evolve charging programs which yield temperatures in agreement with the targets, which holds promise for a practical application of the method in the steel plant.