pended solids such as mineral slurries. Pressure-driven mem-A series of well-controlled membrane filtration experiments are brane filtration can operate at either crossflow or dead-end performed to systematically investigate the dynamic behavior of flow configuration. It is generally accepted that separation permeate flux in crossflow membrane filtration of colloidal suspenof colloids and suspended matter in a crossflow mode is sions. Results are analyzed by a transient permeate flux model advantageous over dead-end operation. which includes an approximate closed-form analytical expression During crossflow membrane filtration, suspended particles for the change of permeate flux with time. The model is based on are transported to the membrane surface by permeate flow a simplified particle mass balance for the early stages of crossflow due to the imposed pressure drop. Because of the finite size filtration before a steady-state flux is attained, and Happel's cell of colloidal particles, particle concentration on the memmodel for the hydraulic resistance of the formed particle cake layer.
The filtration experiments demonstrate that permeate flux brane surface reaches its maximum value after a short period declines faster with increasing feed particle concentration and of time, and a cake layer starts to form (3, 4). Cake layer transmembrane pressure and with a decrease in the particle size formation in membrane filtration can be viewed as a special of the suspension. It is also shown that crossflow velocity (shear case of the more generalized phenomenon of concentration rate) has no effect on permeate flux at the transient stages of polarization (3). The resulting cake layer on the membrane crossflow filtration. Pressure relaxation experiments indicate that surface increases the hydraulic resistance to permeate flow the particle cake layer is reversible, implying no irreversible deposiand, thus, reduces permeate flow through the membrane. tion (attachment) of particles onto the membrane surface or the Understanding the mechanisms of permeate flux decline and accumulated (retained) particles. The experimental results are development of predictive models for the behavior of permeshown to be in very good agreement with the theoretical predicate flux in crossflow membrane filtration are therefore of tions, thus verifying the validity of the model for the transient paramount importance to the design and successful applicapermeate flux in crossflow filtration and the underlying assumptions in the derivation of the model. แญง 1997 Academic Press tions of this technology.