installation. Nonetheless, despite the high cost of such membranes, using them is economical due to their long life. In this study nano pore size plate ceramic membranes were manufactured by a modified gel-casting method using ?-alumina. Natural gelatin, traditionally intended to food industry, was used as the gelling agent, chosen on the basis of its low cost and environmental impact. The effects of important parameters such as sintering temperature, solid loading, powder size and gelling agent content on the porosity and mean pore size of membrane were investigated when ceramic support was fabricated. New method of using liquid desiccant was used for drying of the casted supports. At this step of the work, the effects of concentration and molecular weight of PEG on the time of drying was evaluated. To improve the separation properties of the synthesized supports, they were coated by additional fine alumina layers by dip-coating method. The experiments were designed by Taguchi method and the quantitative effects of the important parameters such as number of layers, solid loading, dipping time, and withdrawal speed of the substrate on the separation performance of membranes were studied by separation of PEG 10000 from its aqueous solutions. The results showed thatthe porosity of the supports increased by lowering the sintering temperature and increasing diameter of the alumina powder.Also, the drying time of the ceramic supports decreased increase in the concentration and molecular weight of PEG used in the desiccant liquid. The most proper supports were obtained when they were sintered at 1400 ? C by using 40 wt % solid and 5 wt % gelling agent in the casting suspension. By conducting the separation tests, the Taguchi method predicted the maximum reachable rejection and permeation flux to be 0.98 and 4.02 (m 3 /m 2 .s ×10 6 ), respectively. According to the predictions, the maximum rejection can be obtained for a membrane coated three times by dipping for 30 sec in a 25 wt % suspension with withdrawal speed of 72 mm/min. On the other hand, the optimized flux can be reached when the support is coated only once by using again a 25 wt % suspension but with different dipping time and withdrawal speed of 20 sec and 36 mm/min, respectively. However, an operationally acceptable membrane, with rejection of higher than 0.9, was obtained when the support was coated only with one layer by using a 25 wt% suspension. In this case the dipping time was 30 sec and the withdrawal speed was adjusted at 36 mm/min. This membrane had the highest flux among all the synthesized membrane according to the Taguchi experimental design. Finally, the molecular weight cut-off of the membranes was found to be 10 KDa which its correspondence average pore size was approximated to be 6 nano meter.
