Study of the prediction model of water flux through a forward osmosis membrane

This work aims at evaluating the accuracy of Tiraferri's model introduced in 2013 for predicting water flux through the forward osmosis (FO) membrane. To this end, a database of FO membranes with their intrinsic parameters and experimental water fluxes was thus constituted. The model was solved numerically in Python Software, first by considering the contribution of external concentration polarization (ECP), and by neglecting it. Using the pressure retarded osmosis mode model, the mass transfer coefficient was adjusted to fit the experimental data to the transport equation. The predicted water fluxes are mostly in agreement with the experimental data, with a resulting mean absolute error (MAE) of 9.18%. This study also shows that the error caused by ECP is less than 1% when deionized water is used as feed solution in FO mode. Because of the reverse salt flux, the ECP was found higher when membranes with high water or salt permeability are tested. The error in van't Hoff prediction, from which the model is based, was found to be low when the draw solution has a concentration in the range of 1 to 2.3 M. Some of the studied membranes exhibit however high MAE, possibly explained by errors from van't Hoff prediction or errors in the membrane parameters. This latter source of error was not accounted for in this study.