Maximizing Cyclone Efficiency: Innovating Body Rotation for Silica Particle Separation via RSM and ANNs Modeling

Abstract

The harmful particles can cause significant health risks and need to be carefully removed from the air before they can pose a threat. One of the most effective methods for separating these particles is a cyclone separator, which can quickly and efficiently remove hazardous particles from the air. In this research, the separation of silica particles using a cyclone separator was investigated and analyzed using artificial neural networks (ANNs) and response surface methodology. The influence of process parameters, including flow rate, particle size, and speed, on cyclone efficiency was investigated. The cyclone experiments were carried out with varying rotation speeds ranging from 0 to 1900 rpm, as well as different particle sizes (15, 25, and 40 μm) and flow rates (30, 50, and 70 m³/hr). Based on the research findings, it was discovered that the ANN model that utilized the multilayer perceptron (MLP) algorithm outperformed the one that used the radial basis function (RBF) algorithm. The findings showed that a neural network with a multilayer perceptron (MLP) architecture performed well in predicting efficiency. Specifically, this MLP had one hidden layer consisting of 10 neurons, and its topology was defined as 3-10-1. The accuracy of the efficiency predictions was high, with a coefficient of determination of 0.998. After analyzing the results, it was concluded that the perceptron multilayer (MLP) model had the highest coefficient of determination value of 0.998 and the lowest error values, with a mean square error of 0.00033838.