Performance Optimization Of A Coal Preparation Plant Using Genetic Algorithms

Modern coal processing plants utilize multiple cleaning circuits to efficiently beneficiate different size coal fractions of run-of-mine coal. An older plant optimization approach is to maintain average product quality from individual cleaning circuits at the same level as the given product specifications for the overall plant. However, many past studies indicate that the equalization of product quality approach fails to produce the maximum plant-yield. A newer approach, known as equalization of incremental product quality, requires that the incremental product quality obtained from each circuit be maintained at the same level to satisfy the overall plant product quality. This approach ensures the maximization of plant yield while satisfying a single product quality constraint. However, while dealing with multiple product quality constraints, this approach by itself may not be sufficient to obtain the desired maximum yield. For the simple reason that the dirtiest particle (or group of particles) with respect to one quality constraint may not be the same particle (or group of particles) with respect to another quality constraint, the mass yield versus product quality relationship generated by equalizing different incremental product quality may not be exactly the same. Therefore, an additional search technique has to be utilized to determine the global-maximum value of plant yield. Thus, the main objective of this study was to utilize an emerging optimization technique, known as genetic algorithms (GA) to maximize plant yield while satisfying multiple product quality constraints. The maximum plant yield obtained from this approach was nearly same as the maximum yield obtained by the incremental product quality approach while satisfying one specific product quality constraint. The GA was applied on a coal preparation plant that utilizes four circuit operations - heavy medium bath, heavy medium cyclone, spiral and froth flotation. The results showed that using GA as an optimization process gives 2.56% higher yield that will result in additional revenue generation of $5,120,000 per annum than average product quality approach.