Experimental study and artificial intelligence modeling of liquid-liquid mass transfer in multiple-ring microchannels

Abstract

This paper reports the results of using multiple-ring microchannels for enhancing liquid-liquid extraction performance. The effects of geometrical parameters including ring and distance characteristics on the extraction efficiency were studied. The mass transfer performance was analyzed using Water + Alizarin Red S+1-octanol system. By change in geometrical parameters, the extraction efficiency of multiple-ring microchannels improved up to 62.9% compared with that of the plain one. The performance ratio is defined based on two contrary effects of friction factor and extraction efficiency for evaluating the extraction performance. A performance ratio of 1.5 was achieved that confirmed the advantage of using this type of microfluidic extraction system. Artificial neural network and adaptive neuro-fuzzy inference system were utilized to evaluate the performance ratio of the multiple-ring microchannels. The mean relative error values of the testing data were 0.397% and 0.888% for the neural network and the neuro-fuzzy system, respectively. The estimation accuracy for both models is appropriate, but the precision of the neural network id higher than that of the neuro-fuzzy system. The genetic algorithm approach was employed to develop a new empirical correlation for predicting the performance ratio with a mean relative error of 1.558%.