Monitoring furnace sidewall integrity using multivariate statistical models

Abstract

Sidewall accretion in a smelting furnace plays an important role in furnace integrity as it provides protection to the sidewall cooling elements and thereby extends the furnace campaign life. The extreme thermal and mechanical stresses in the furnace and changes in the slag chemistry can cause the sidewall accretion to melt or break away decreasing the lining protecting the cooling elements. Thin accretion is highly undesirable as it increases the risk of exposing the cooling elements to the molten bath and in the worst case may lead to a failure. Therefore, monitoring the sidewall accretion on an operating furnace is extremely important to manage furnace integrity and to support safe furnace operation. Operators often generate estimates for sidewall accretion and refractory thickness using either copper cooler temperatures or heat load into the cooling water circuits. Multivariate statistical models, on the other hand, have the capability to consider all the process measurements simultaneously to provide a more reliable estimate. This paper describes the application of multivariate statistical models for sidewall integrity monitoring on a Flash Converting Furnace (FCF) and also discusses the online implementation of these models as the Sidewall Accretion Monitoring System.