Artificial neural network model for turbulence promoter-assisted crossflow microfiltration of particulate suspensions

Abstract

In this study, an artificial neural network (ANN) model for the turbulence promoter-assisted crossflow microfiltration (CFMF) process was successfully established, in which the inlet velocity, transmembrane pressure (TMP) and feed concentration were taken as inputs, and the flux improvement efficiency (FIE) by turbulence promoter was taken as output. Using the trained ANN model, the FIE can be predicted under CFMF operation conditions that are not included in the training database. It reveals that the FIE first increases and then decreases with increasing either TMP or inlet velocity, and increases with increasing feed concentration. Among three input variables, TMP has the most important effect on the FIE. The optimization of MF operation conditions was largely dependent on the feed concentration. The high FIE can be obtained by exerting both high inlet velocity (>0.7m/s) and low TMP (<30kPa) at a relatively low feed concentration (<1g/L), and both high inlet velocity (>0.7m/s) and high TMP (>70kPa) at a relatively high feed concentration (>8g/L). This study provides a useful guide for the applications of turbulence promoter in CFMF processes.