MULTIPLE LINEAR REGRESSION METHOD UTILIZATION TO THE BLAST FURNACE PELLETS COLD COMPRESSION STRENGH ANALYSIS

The blast furnace pellets produced with hydrated cal present a high mechanical embrittlement trend mainly connected to the cold compression strength parameter (CCS). The mechanical performance enhancement of them is associated to a distinct group of parameters interaction. During 2010's year a relevant drop was found in the high basicity blast furnace pellets physic quality, particularly in the cold compression strength parameter. The search for the main causes of this anomaly was developed using statistic tools as the analysis of variance (Anova) and the multiple linear regression method (Best Subsets e Stepwise Regression). These methodology of analysis employment bare the pellets production critical process parameters where were regarded the quite relevant contribution of the grinded iron ore size, the binary basicity and the carbon content to the final product quality improvement. The created multiple linear regression model presented an adjusted R2 of 67%, that made possible the simulations to forecast the cold compression strength behavior. The analysis results provided information to the problem lessening, which achieved an average quality gain of 20 daN/p related to the cold compression strength of the pellets.