Reverse osmosis separations of multicomponent electrolyte solutions

A new model was developed to treat reverse osmosis separations of electrolyte solutions. The model consists of a system of transport equations based on the Extended Nernst-Planck equation, which includes diffusion, convention, and electromigration. Boundary conditions include a distribution coefficient that is due to specific interaction potential representing repulsion of ions from the membrane material. Boundary conditions also include potential jumps known as Donnan Potentials. The model incorporates a mechanism for a varying membrane fixed charge as a function of ion concentrations and pH inside the membrane. In addition to salt ions, hydrogen and hydroxide ions are also considered, as pH changes indicate they take part in the transport.; A reverse osmosis system was constructed for gathering data necessary to test the model derived in this work. The membrane module was constructed to allow for high pressure membrane potential measurements. The membrane material was characterized, and a variety of analytical techniques were used to determine measured values, which include species concentrations, pH, and membrane potential.; A variety of single salt experiments were performed over a wide range of concentration and pH using NaCl and CaCl{dollar}sb2{dollar}. The model was shown to be quite effective in producing the same type of separations behavior as was experimentally observed. Using a single set of parameters, the model was able to fit rejection behavior quantitatively well over the experimentally accessible range of concentration and pH. Predictions of {dollar}Delta{dollar}pH and membrane potential were qualitatively good for the same experiments. The model parameters determined from the data were physically reasonable.; Using the parameters obtained from single salt experiments, separations of mixtures of NaCl and CaCl{dollar}sb2{dollar} were predicted and experimentally confirmed. Agreement between the predictions and experimental data was quite good. The model can additionally be used to simulate conditions beyond those used to obtain experimental data.