Fluid flow in hydrocyclones optimized through multi-objective genetic algorithms

Abstract

Transient swirl flow fields that exist in the hydrocyclone separators have been simulated using the CFD software FLUENT™. The outputs are treated in a multi-objective fashion using tailor-made evolutionary computing softwares developed in-house. The recently proposed ‘new multi-objective genetic algorithm’ developed at this research group is utilized for this purpose along with a multi-objective immune system algorithm. Using the simulated flow fields, attempts are made to simultaneously optimize two conflicting criteria: (i) the volume of the LZV (locus of zero vertical velocity) envelope that governs the extent of classification towards the overflow region and (ii) the overall pressure drop that drives the classification process towards the underflow. The resulting Pareto frontiers are computed and analysed.

Keywords:

• flow configuration
• hydrocyclone
• NMGA
• Pareto frontiers
• CFD
• genetic algorithms
• multi-objective optimization

Acknowledgements

The authors are thankful to TATA Steel for funding this research, to Prof. G.S. Dulikravich for his generous invitation and financial support during the IPDO 2007 conference and to B. Siva Kumar and V. Satish Babu for their preliminary work on NMGA.