Computational model for estimation of refractory wear and skull deposition in blast furnace hearth wall

Knowing how much refractory remains in the hearth is critical to the assessing when a blast furnace hearth needs to be relined. In this work a computational model coupled with a finite state machine and a neural network pattern recognition block has been developed for the blast furnace hearth to determine the thickness of two refractory layers and formation of protective layer of solidified metal (skull). A neural network was also used for data correction. The results provide estimation of wear of the hearth refractory lining and insight to the erosion profile formed inside the blast furnace hearth. The walls and the floor of the hearth have embedded thermocouples to monitor the temperatures of the furnace walls. Based on the temperature readings of the thermocouples one can determine the heat flux through the wall. This heat flux is used in the computational model, based on heat flow and conservation of energy, to determine the skull deposition and refractory wear.