Hybrid ethanol dehydration systems are modeled and optimized using MINLP. The systems consist of a distillation column for approaching the ethanol/water azeotrope and of a pervaporation unit for producing pure ethanol. The optimal design and operating parameters including number of trays, feed location, reflux ratio, number of membrane sections in series and the number of membrane modules in each section are determined. Regression equations fitted to solutions of differential equations are employed for modeling the membrane modules. Quadratic and exponential regression, as well as metric and linear interpolation are studied for approximating the integral membrane model; the exponential approximation is selected. A new mathematical representation of the superstructure of the membrane subsystem is suggested and applied. A successive refinement method with non-increasing number of binary variables is developed and successfully applied. Computational experiences with GAMS/DICOPT are presented. Using our new membrane superstructure representation, the hybrid system can be optimized effectively. The optimization method developed is also successfully applied for process intensification of an industrial scale dehydration plant. Compared to the existing plant, 12% savings in the total annual cost can be achieved by applying 32% additional membrane surface, in consequence of a radical decrease in the reflux ratio (3.3:1.4) in the column, and of producing less concentrated alcohol in the distillate. Sensitivity of the total annual cost to the specified ethanol yield, overall membrane surface and membrane replacement cost is studied. Total permeate recycling is found to be more economical, compared to partial recycling.