Modeling and analysis of blast furnace drainage

The blast furnace is the prevailing unit process for iron-making in the steelmaking process and the present growing global demand for steel leads to higher requirements set on the efficiency and productivity of the process. The lower part of the blast furnace, where iron and slag accumulate, the hearth, is key to the furnace performance both from short-term and long-term perspectives, so a monitoring of the hearth state is a critical task for a successful operation. However, it is difficult to provide a detailed description of the processes due to the complex dynamic phenomena that take place in the hearth, beside the extreme chemical and thermal conditions that prevent or at least strongly limit the access to direct measurements. As consequence, the hearth state is for the most part unknown.

The models presented in this thesis aim to shed some light on the hearth conditions and the drainage of iron and slag based on estimates of the production rates of iron and slag, measurements of the outflow rates of iron and slag, and other process measurements interpreted by mathematical models. Drainage data available at a large furnace with three tapholes was used in the analysis. The observed draining rates were filtered and systematically organized by a method based on principal component analysis, which revealed interesting outflow patterns and evolutions of these in time. On the basis of the results, some assumptions regarding the hearth state, primarily about the dead man and the taphole, were evaluated in two off-line models that were developed to estimate the iron and slag levels in the hearth. The approaches of these off-line models were a two-pool hearth with liquid outflows expressed by a parameterized expression inspired by measurements in the plant, and a single-pool hearth with a physical description of the conditions at and in the taphole.

To be able to provide a real-time view of the conditions, an on-line model of a multi-pool hearth system was developed based on inflow and outflow estimates continuously obtained from the plant to reconstruct the liquid levels.

The sensitivity of the models to changes in their parameters was studied and the results were appropriately compared with observations from the reference furnace. Fairly good correlations between the calculated and observed variables were found, gaining confidence in some of the assumptions made in the modeling. Particular focus was put on analyzing the role of coke permeability at the taphole and the taphole conditions. The fluctuations and drifting seen in the process data indicates a high level of complexity of the system at hand. Despite these challenges, the models developed are deemed to have a potential of being applied as tools to assess different hearth states and the system’s response to certain conditions. Furthermore, the on-line model can be applied to perform the crucial task of estimating and tracking the liquid levels in the hearth. Future work will be focused on extending and integrating some of the features of the models to yield a better description of the hearth state. Furthermore, an adaptation of the off-line models should be studied to track changing states of the blast furnace hearth.