Estimation of diameter and surface area flux of bubbles based on operational gas dispersion parameters by using regression and ANFIS

Adaptive neuro fuzzy inference system (ANFIS) procedure and regression methods were used to predict the Sauter mean bubble (bubble diameter) and surface area flux of the bubble in a flotation process. The operational conditions of flotation, impeller peripheral speed, superficial gas velocity, and weight percent solids were used as inputs of methods. By using the mentioned operational conditions, the non linear regression results showed that Sauter mean, and surface area flux of the bubble are predictable variables, where the coefficients of determination (R2) are 0.57 and 0.74, respectively. To increase the accuracy of prediction an ANFIS model with cluster radius of 0.4 was applied. ANFIS model was capable of estimating both Sauter mean, and surface area flux of the bubble, where in a testing stage, satisfactory correlations, R2 = 0.78, and 0.86, were achieved for Sauter mean, and surface area flux of bubble, respectively. Results show that the proposed ANFIS model can accurately estimate outputs and be used in order to predict the parameters without having to conduct the new experiments in a laboratory.