The Comparison of Fuzzy Inference System and the Bees Algorithm for Optimal Design of Balancing Hole of Centrifugal Pump Impeller

Abstract

This study presents a comprehensive comparison of two distinct computational approaches, Fuzzy Inference System (FIS) and The Bees Algorithm (BA) for optimizing the design of balancing holes in centrifugal pump impellers. These holes are essential for reducing axial thrust without compromising hydraulic efficiency. The optimization parameters include hole center angle, hole diameter, radial placement, and the number of holes. By employing computational fluid dynamics (CFD) simulations, a data-driven regression model was derived from 111 configurations of a TKF 125-400 pump model. The Mamdani-type FIS uses triangular and trapezoidal membership functions and linguistic rules, while the BA, inspired by the foraging behavior of honeybees, utilizes a swarm intelligence approach for multi-objective optimization. Comparative performance is evaluated based on axial thrust minimization and efficiency retention. The findings suggest that while BA provides superior optimization accuracy, the FIS offers greater speed, interpretability, and adaptability for real-time applications in pump design.