Analysis and optimization of louvered separator using genetic algorithm and artificial neural network

Abstract

Louvered separators are widely used in many engineering applications. This study aims to optimize four geometrical parameters: the blade length (Lb), the gap between blades (Hg), and the dust container width (Ld) and height (Hd). The objective functions were the Euler number and cut-off diameter. First, a direct multi-objective optimization study was performed using the CFD results. Then, a surrogate-based approach using the radial basis function artificial neural network for single and multi-objective optimization was employed. Increasing the height of the collector and the gap distance between the blades increased the pressure drop, while increasing the blade length and the collector length decreased the pressure drop. Furthermore, increasing the height of the collector and the blade length decreased the separator efficiency, while increasing the gap between the blades and the collector length increased the efficiency. The optimized cut-off diameter was 1.663 μm, and the minimized Euler number, Eu, was 21.03.