Prediction of trapping efficiency of vortex tube ejector

Abstract Vortex tube ejector is employed to extract sediments from canal. It consists of a duct laid across whole bed of the canal with a slit along its top edge and compared to the other alternative sediment-extraction devices, it is very efficient and economical. In this study, adaptive neuro-fuzzy inference system (ANFIS), artificial neural network (ANN) approaches were employed to predict the trapping efficiency of vortex tube ejector. Data-set as many as 144 was obtained by conducting experiments on vortex ejector. Out of 144 data-set, 100 data selected randomly were used for training whereas remaining 44 were used for testing the models. Input data-set consists of sediment size (mm), concentration of sediment (ppm), ratio of slit thickness and diameter of tube, (t/d) and extraction ratio (%) whereas trapping efficiency (%) was considered as output. Three membership’s functions, i.e. triangular, generalized bell-shaped, and Gaussian were used with ANFIS. A comparison of results suggests that Gaussian membership function-based ANFIS model performs well in comparison to other membership functions-based ANFIS models, ANN and predictive equations proposed by previous researchers. Sensitivity analyses suggest that extraction ratio is the most important parameter in estimating trapping efficiency of vortex ejector.