Dynamics of a flash smelter

Abstract

Norddeutsche Affinerie is running an Outokumpu type flash smelter for more than 25 years. During the last years the concentrates smelting capacity was increased to 750, 000 tpy, which is double the initial design capacity. Maintaining this high furnace performance level requires very good metallurgical process control. It is simply not enough just to base this control on thermodynamic and kinetic model since these neglect the dynamics of the furnace. These dynamics originate due to non-stationary operation affected by for example changing concentrates and feed mixtures (poorly defined return slags and flue dusts), or changing heat transfer due to accretions, etc. Understanding these dynamics and being able to quantify these can have significant financial benefits. In this paper a mass and energy balance model is developed and discussed based on measured data using data reconciliation. Presently this time the model is based on the major elements copper, iron, sulfur and silica. With this model it is possible to quantify the differences of the various material streams such as matte, slag, flue dust, matte carry-over to settling furnace, return slags and flue dusts, etc. The calculated data show the evolution of the dynamics of the process. These dynamics are related to various input parameters using methods such as statistics and (time lagged recurrent) neural nets.