Steady-state simulation and optimization of gravity separation circuits by use of linear programming and expert systems
A knowledge-based system for the steady-state simulation and optimization of gravity separation circuits is presented. The only data required by the simulation models are the number of separation banks, number of species, feed rates, composition of feed and the separation factors of the different species. A knowledge-base which permits the intelligent choice of bounds, is used to generate the bounds on the separation factors. The knowledge-base may be updated by the user or the simulation model itself. In this paper, a gravity concentration plant is described by two linear models, one being the subset of the other. The computer program formulates and solves these linear models sequentially by using any suitable linear programming (LP) package. The first model optimizes the flow rates and the path of the valuable elements between the separator banks. This model produces the optimal flow rates of the valuable elements between the banks, as well as the circuit configuration. The second model simulates the flow rates of all elements in such a way that the concentrate grade is maximized. An optical grade for the valuable elements in the concentrate is given by the second model. The circuit configurations, grades and recoveries produced by the simulation model are similiar to those encountered in industry. This model may also be used to suggest an alternative circuit structure and operating conditions in the case of drastic changes in the operating environment of the plant. The simulation model is very time efficient and produces solutions within minutes. © 1991.