Statistical analysis and prediction of Al 2 O 3 –SiC–C based blast furnace trough refractory wear rate

Abstract

In the present study, statistical analysis is carried out to quantify the blast furnace trough (BFT) refractory wear. To this end, two key parameters, namely linear (mm/s) and material (kg/s) wear rates, are identified to characterise refractory wear in terms of hydrodynamic, thermal, chemical, and structural parameters known to govern the BFT wear rate in practice collectively. Linear wear is defined as the maximum decrease in the working layer thickness at the end of a BFT campaign, while material wear is defined as the amount of material used to re-line the BFT at the end of a campaign. Initial analysis indicated that a simple multiple linear regression model does not accurately predict observed refractory wear, possibly due to complex and non-linear interactions between independent and dependent variables. Consequently, a regression-based second-order polynomial model embodying industrial BFT wear rate data has been developed. Results thus obtained have been assessed against a separate set of validation data, and this confirmed that the second-order polynomial model of linear wear rate can satisfactorily predict the wear rate of BFT refractory within ±10%.