| The purpose of this research was to develop a membrane module that offers an improved mass transfer coefficient over existing modules by using hollow fiber membranes positioned transverse to the liquid flow path. This arrangement is well known to give better transfer coefficients than parallel flow devices. Transverse flow was accomplished by weaving hollow fiber membranes into a fabric that was sealed into and wrapped spirally around a central permeate tube. In this work, the well-known sorption-diffusion model was used to describe pervaporation, and a resistance-in-series relationship was derived to describe mass transfer.; The model substance for most work was O{dollar}sb2{dollar}, so the permeability of the membrane of choice, PDMS, for O{dollar}sb2{dollar} was determined experimentally. Two mini-modules were made with fibers parallel to one another and liquid was pumped inside the fibers. The transfer coefficient in this arrangement is known to vary with velocity to the 1/3 power, and a plot of 1/Ko vs. (L/v){dollar}spwedge{dollar}1/3 was reasonably linear. The intercept corresponded to the overall mass transfer resistance at infinite velocity and, therefore, represented the membrane resistance. From this, the permeability was calculated to be {dollar}2.5times10sp{lcub}-8{rcub}{dollar} mol/(m s atm), which is 30% higher than the oft-quoted literature value of Robb (40), though the difference is not statistically significant. It is worth noting that if this permeability value is inaccurately high, our reported results for k are inaccurately low. Therefore the permeability value used is conservative.; Small fabrics were woven with silicone rubber hollow fiber membranes and monofilament nylon. Batch experiments yielded a significant variation in mass transfer coefficient, k, for the different fabrics tested. Also, {dollar}Delta{dollar}P/L was measured for each fabric, and a wide variation was observed. An optimal fabric was determined based on high mass transfer coefficient, low pressure drop, and high membrane packing density. Tests were conducted involving both fabric and flat sheet membrane, and the flat sheet did not contribute to mass transfer in proportion to the space it would take up in a module.; Two larger prototype modules were made, each with a fabric 22 cm x 35 cm wrapped in 4 layers around a central permeate tube. For O{dollar}sb2{dollar} removal, observed k values for the prototypes were slightly below those observed in small-scale tests. Values of k near (and one above) 0.01 cm/s were observed, which is excellent. Tightly wrapping the fabric decreased performance. For removal of trichloroethylene (TCE), some membrane fouling occurred. Presumably, petroleum jelly used on o-rings was dissolved by the TCE and subsequently sorbed into the hollow fibers. Membrane discoloration occurred, which was accompanied by flux reduction.; Observed mass transfer coefficients (during O{dollar}sb2{dollar} removal) and membrane packing densities for the prototype module were considerably higher than those reported for traditional spiral wound modules. The mass transfer coefficients were higher than those of conventional hollow fiber modules. |
