渗透蒸发过程在有机溶剂（醇，苯，丙酮等）的脱水上有着很大的潜在应用前景，该种膜过程在国外已被GFT公司工业化，用于无水已醇的制备，但在国内该过程还只局限于实验室研究；这在一定程度上与传质机理的研究不够深入有关。 现有的平衡溶解扩散模型在解释渗透蒸发实验时还不够令人满意，根据该模型，渗透通量与膜厚成线性反比，分离因子与膜厚无关，而这与实际情况是不符合的，因此，本文提出了非平衡溶解扩散模型。 非平衡溶解扩散模型认为，渗透通量（Ji，Jj）和分离因子α与下列几个参数有关：膜上游表面的平衡浓度值和实际浓度值（Cie，Cje，Ci，Cj），组分向膜中溶解时的溶解速度系数（Ki，Kj），组分的平均扩散系数（Di，Dj）以及膜厚δ, 用下式表示： Ji=Ki(Cie－Ci)=Di Ci/δ Jj=Kj(Cje－Cj)=Dj Cj/δ α=(Ji/Jj)/(Xi/Xj) 本文提出了一种测定组分向膜中溶解时溶解速度系数的方法。并对测量的理论基础进行了探讨，而且实际测量了35℃下85%乙醇水溶液向PVA膜中溶解时的溶解速度系数。 本文还提出了使用平均扩散系数以简化对膜内扩散过程的描述，并实测了水和乙醇在PVA膜中的平均扩散系数。 欲测定以上各位，需快速准确地知道组分在膜内的浓度，为此，本文还提出了一种新的测定醇水在PVA膜中溶解度的方法。 根据以上非平衡溶解扩散模型，对渗透蒸发进行了摸拟计算，结果表明，分离因子随膜厚减小而增大；渗透通量虽随膜厚减小而增大，但并非线性反比关系。 本文还利用上述提到的测定膜内溶质组成的方法，研究了醇水在PVA膜中平衡溶解度和平衡溶解组成与渗透蒸发性能之间的关系，得到如下结论：①渗透通量的变化与平衡溶解度的变化趋势一致；②渗透液中醇的含量与醇在膜中平衡溶解组成变化趋势一致。 这两点结论与非平衡溶解扩散模型的结论对制膜有一定的指导作用： （1）在对膜进行处理（热处理和交联等）时，应使处理后（i）组分在膜中溶解度和扩散系数尽可能大，以保证大的通量；（ii）各组分向膜中溶解时溶解速度系数的差异要尽可能大，以保证大的分离因子。 （2）选择适当的膜厚，不必刻意追求过薄的膜，以保证制膜时不易出现缺陷，同时也能获得较大的通量和分离因子
Pervaporation has wide potential application in the dehydrationof organic solvents such as alcohol, benzene, acetone and so on.The membrane process has already been industralized in theproduction of absolute ethanol by GFT Company. Pervaporationhas also been widely studied in China, But up to now, themembrane process has just been studied in laboratoty scale. Thisis partially due to the inadequate study on mass transfer process. The existing mechanism - the equilibrium solution diffusionmodel - is not satisfactory. Calculations based on this modelshow that the flux is linearly inverse proportional to membranethickness and the separation factor is independent up on membraethickness, which doesn't agree with the actual experiment results.Under the circumstances, a non-equilibrium-solution diffusionmodel is proposed. According to the non-equilibrium model, the fluxes(Ji, Jj) andseparation factor(α) depend upon the fluxes (Ji, Jj) and equilibrium and actual concentrations of components at theupstream membrane surface(Cie, Cje, Ci, Cj), the solution ratecoefficients(Ki, Kj), the average diffusion coefficients(Di, Djand membrane thickness(δ). Ji = Ki (Cie-Ci) = Di Ci/ δ Jj = Kj (Cje-Cj) = Dj Cj /δ α = (Ji/Jj)/(Xi/Xj) In this non-equilirium model, we need to know the solution ratecoefficients of the components (in our experiment, water andethanol) into PVA membranes. Therefore, an experiment method isdeveloped and the theoretical basis of the method is analyzed.The solution rate coefficients of water and ethanol solutinginto PVA membranes are dertermined at 35℃ by measuring the swelling xurves of PVA membranes in 85% ethanol solution. The average diffusion coefficients of water and ethanol insidePVA membranes are measured. The using of average diffusioncoefficients simplifies the description of the diffusion processof water and ethanol inside the membranes, In the studies mentioned above, we should know the concentrationsof water and ethanol inside PVA membranes, Therefore we havedeveloped a method to determine the compositions of eachcomponent inside the membranes by gas chromatographer. The simulation of pervaporation based on the non- equilibrium-solution diffusion model shows that the separation factorincreases with the decrease of membrane thickness, and that theflux isn't linearly inverse proportional to membrane thickness,although it does increase with the decrease of thickness. Comparing the data of equilibrium solution properties of waterand ethanol inside PVA membranes and the data of pervaporationexperiments, the relationship between equilibrium solution andpervaporation can be found as follows:(i) pervaporation flux andequilibrium solubility have similar changing tendency; (ii)equilibrium solution composition and permeate composition havesimilar changing tendency, These two conclusions and the conclusions drew from non-equilibrium-solution diffusion model can be taken as guides inmembrane preparation: (1) the membrane should be accordingly treated(heat-treated,crosslinked,etc.) so that:(i) the equilibrium soluiblities andthe diffusion coefficients should be large in order to obtainlarge permeate flux; and (ii) the difference of solution ratecoefficients between the two components should be large so as toobtain large separation factor; (2) Appropriate membrane thickness(δ) should be chosen to getsatisfactory flux and separation factor. Pursuing a membrane ofUnnecessary thinness would cause increasing defects (e.g. cracksand pin holes) in membrane preparation.
