Circuit analysis: optimizing mineral processing flowsheet layouts and steady state control specifications

Abstract Process optimization aims to extend flowsheet simulation technology with techniques for determining plant control strategies and, ideally, circuit structural layout that result in optimum performance of the circuit in economic terms. Such an endeavor is confronted with problems of model scale, nonlinearity, and the general difficulties of mixed discrete/continuous optimization. This paper shows how the structure of the flowsheet models for mineral processing circuits facilitates overcoming these difficulties in many applications. Specifically, the same structural features used in the development of circuit analysis (Meloy, 1983) are employed here with the addition that a formal probabilistic interpretation is given to the unit partition coefficients. Original optimization algorithms, extensions of the techniques of stochastic dynamic programming, are reported. These algorithms have capability to determine control settings for the unit operations that maximize the net value of the circuit yield. Moreover, they can select the best of a number of discrete circuit layout options posed by the system designer, considering fully the scope of control adjustments for all candidate vessels in the system. The algorithms are computationally efficient and are readily applied to full-scale models. Several issues relevant to application of the algorithms are discussed.