The Use of Models in the Design and Operation of Metallurgical Plants

In the technical sense, as well as in the professional sense, the word 'model' can have many meanings. More than 23 000 references were uncovered by a simple literature search on keywords ' MO D EL#. and. (META LLURG#. or. M I NERAL#)' . For the purpose of this paper, 'models' will be considered to be mathematical formulations which attempt to represent or approximate a process. The model may be purely, or a combination of: ò fundamental models, based on the laws of physics, chemistry and thermodynamics; ò semi-empirical or phenomenological models, where an intuitive concept of the process is formulated and found by testwork to fit observed behaviour; and ò empirical or 'black-box' models (including neural networks), where a mathematically convenient form is proposed and fitted. Even under this more restricted definition, which eliminates heuristic and statistical 'models', it is clear that the efforts expended in devising and using models have exploded during the years since the first APCOM Conference in 1961, benefiting from the dramatic decrease in the cost of computer power. This paper reviews the areas of usage of computer modelling in the metallurgical and mineral processing industries, from plant design, through off-line support, to on-line modelling and automatic control. In each area, frontiers of new development of the technology are mentioned. It is noted that, while modelling is now standard practice in design, it is not yet the norm in operating and controlling plants. The reasons for this partial acceptance are, it is argued, less related to the limitations of the technology than to the human and organisational factors involved in transferring technology to implementation.