(658b) The Importance of a Well-Designed Cross-Flow System In the Removal Study of Adsorbing Estrogens by Nanofiltration Membranes

At laboratory scale, cross-flow systems with a slit channel have been widely used in the research of membrane water treatment, particularly when the focus lies on removal of trace contaminants. Slits constitute a laboratory model of spiral-wound modules allowing for the study of the fundamentals of mass and momentum transport mechanisms in the feed channel by controlling its hydrodynamics. Spiral-wound modules are, in turn, the most common membrane geometry used in water treatment, due to their compact geometry with high surface/volume ratio.

This work aims at demonstrating the importance of having a well-designed cross-flow system and without oil leakages from the pumping system in the removal of endocrine disruptors by showing how both an ill-designed cross-flow system and the presence of oil in the aqueous solution impact on the results of their removal by NF membranes. For that, a comparison between hormone filtration results yielded by both type of systems is performed. Additionally, further experiments performed with the well-designed system aim at defining, understanding and quantifying the physical mechanisms of momentum and mass transfer controlling the adsorption of hormones onto polymeric NF membranes, viewing the development of a predictive model.

In an ill-designed system, pump delivering flow rates to the cross-flow membrane cell above a limit value (Reynolds number based on the hydraulic slit diameter, Re_dh, above 10,000) causes the membrane active layer to be ripped off from the support layer, and yields a E2 retention below 15% whereas that obtained for a Re_dh number of 2800 in a well-designed cross-flow system is higher than 80%. On the other hand, the presence in the aqueous solution of oil contamination originated from the pump of the ill-designed system caused the E2 retention to be much lower than expected: 25% in the presence of oil against 65% in the absence of oil obtained in the well-designed system. These and other results led to the abandonment of the ill-designed system since, in opposition to it, the well-designed one allowed to mimic spiral-wound hydrodynamic conditions and carry out the work aiming at understanding the fundamentals of the removal mechanisms of hormones by NF membranes. Special care was taken in the new system to re-design the pump head of a commonly used diaphragm pump in cross-flow systems to make sure that there was no oil leakage. These findings are very important for pump manufacturers and users.

Once these changes were made, it was possible to proceed with the experiments viewing the definition of a predictive model of adsorption of hormones onto polymeric NF membranes. First, knowledge from experiments was acquired on how hydrodynamics and operating conditions -  pressure, circulating Reynolds numbers (Re_h) and feed concentration - affect hormone adsorption: adsorption is governed by the initial concentration at the membrane surface that, in turn, depends on the hormone feed concentration, operating Re_h and pressure. Membrane retention, however, depends on the initial polarisation modulus, given by the fraction of the initial concentration at the membrane surface divided by the initial feed concentration. Then, based on these results, a sorption model was developed, predicting very well both feed and permeate transient concentrations for two different hormones in the normal range of operating pressures and Re_h of spiral-wound membrane modules. These findings are an important advancement in determining which membrane would be more suitable to effectively remove hormones since the model predicts well the feed and permeate concentrations with a substantial reduction of experimental work.