GRIDING BALL REPLACEMENT PLANNING PROBLEM: A CASE STUDY

This study addresses the grinding ball replacement planning problem. We proposed a fuzzy controller to determine the recommended power of the mills and a predictive model to estimate this power. Also, we proposed a MILP formulation and an algorithm based on Enhanced Iterated Local Search (E-ILS) to determine the instant and the recommended bulk weight of the grinding balls to be replaced in each mill. We test the methods on instances from the Vale S.A.'s Cauê Plant. Compared to Gurobi, E-ILS achieved optimal solutions with an average variability of 1%. With this good result, we integrate the E-ILS into a decision system with a two-level architecture. The higher level plans the replacement with E-ILS, while the lower level executes it in the industrial PLC. Implementation in 2024 resulted in a 17% reduction in the consumption of grinding balls.