Béatrice Balannec

Treatment of dairy process waters by membrane operations for water reuse and milk constituents concentration

Abstract In the dairy industry non-accidental losses of milk or dairy products to sewer amount to 1–3% of the total milk processed. These significantly contribute to the 0.5–6.0 g COD/L of end-of-pipe wastewater. The major part of this polluting charge originates from starting, interrupting, and stopping dairy plant procedures, where milk-products are diluted with water and discharged to a purification station or collected to be spread on land. The very few works, that have been dedicated to the treatment of the so-called process waters (flushing waters, first rinse waters or “white waters”), show that nanofiltration (NF) or reverse osmosis (RO) is adequate for the concentration of milk components. The present work reports NF and RO performances (permeate flux, milk components rejection) of an effluent model solution (diluted skimmed milk). Performances of eight NF and RO membranes were compared by dead-end filtration. Crossflow experiments with NF and RO spiral-wound membranes confirm the results obtained by dead-end filtration. The results showed that one single membrane operation allowed the milk constituents to be concentrated in the retentate but reusable water of composition complying with the standard of purified water from process water was not reached. A finishing step (RO membrane, other) is needed for the production of reusable water.

Computation of the hindrance factor for the diffusion for nanoconfined ions: molecular dynamics simulations versus continuum-based models

We report the self-diffusion coefficients and hindrance factor for the diffusion of ions into cylindrical hydrophilic silica nanopores (hydrated silica) determined from molecular dynamics (MD) simulations. We make a comparison with the hindered diffusion coefficients used in continuum-based models of nanofiltration (NF). Hindrance factors for diffusion estimated from the macroscopic hydrodynamic theory were found to be in fair quantitative agreement with MD simulations for a protonated pore, but they strongly overestimate diffusion inside a deprotonated pore.

Experimental study and modelization of reverse osmosis with salt solutes in an unstirred batch cell

The study deals with reverse osmosis of salt solutes in an unstirred batch cell at a rejection coefficient very close to 1. The experiments were performed in a stainless steel batch cell specially manufactured in order to lower the dead volume under the membrane. The operating conditions led to salt concentrations at the membrane rising from 1 to more than 100 kg.m−3 during a given experiment. Introducing a reduced number of restrictive assumptions, a model was used in which the convection-diffusion equations associated with the osmotic theory equation are solved by a mixed-finite elements method. Unlike other charged solutes (such as proteins), this model very well describes the experimental results of salt solutes, even if salt concentration near the membrane becomes high.