Micellar enhanced ultrafiltration (MEUF) of mercury-contaminated wastewater: Experimental and artificial neural network modeling

Abstract

MEUF was applied efficiently to remove mercury (Hg) from simulated wastewater by using polyacrylonitrile membrane and sodium dodecyl sulfate (SDS) as surfactant. In this process leakage of surfactant monomer to permeate water causing a secondary pollution that was addressed by using MEUF followed by activated carbon fiber (MEUF-ACF). The effect of operating parameters including, concentration of Hg and pH of feed solution, molar ratio of SDS to Hg, and retentate pressure was investigated to optimize MEUF process. Moreover, artificial neural network (ANN) model was proposed to predict Hg removal efficiency to optimize MEUF process without doing laborious and time-consuming experimental work. ANN model performance was evaluated on the basis of statistical values such as mean square error (MSE) and coefficient of determination (R2).The experimental results presented that optimum operating parameters were found as 10 ppm of Hg concentration, pH 7.0, molar ratio of SDS to Hg 8:1 and retentate pressure was 1.5 bar. MEUF results showed 95.75% and 50.91% removal of Hg and SDS, respectively, while 96.83% Hg and 97.15% of SDS rejection was achieved using MEUF-ACF. The statistical values of proposed ANN model presented high degree of agreement between experimental and predicted values by ANN (R2>0.95 for training, validation and testing dataset). Resultantly, MEUF-ACF can eradicate issue of secondary pollution and proposed ANN model can be a competitive, powerful and fast alternate to laborious experimental work for MEUF process optimization.