BACKGROUND: The design of experiments (DoE) is applied to the process optimization of p-xylene (pX) separation from its isomers m-xylene (mX) and o-xylene (oX) mixture using silicalite-1 membrane supported on alpha-alumina. A central composite design (CCD) coupled with response surface methodology (RSM) was used to correlate the effect of two separation process variables, temperature (150-250 degrees C) and pX feed partial pressure (0.10-0.26 kPa) to three responses: (i) pX flux; (ii) pX/oX separation factor; and (iii) pX/mX separation factor. The significant factors affecting each response were elucidated from the analysis of variance (ANOVA). The interaction between two variables was investigated systematically based on three-dimensional response surface plots. RESULTS: The optimization criteria were used to maximize the value of pX flux, pX/mX separation factor and pX/oX separation factor. The optimum pX flux of 5.94 x 10(-6) mol m(-2) s(-1), pX/oX separation factor of 19 and pX/mX separation factor of 20 were obtained at a temperature of 198 degrees C and pX feed partial pressure of 0.22 kPa. CONCLUSIONS: The experimental results were in good agreement with the simulated values obtained from the proposed models, with an average error of +/- 2.90%. In comparison with the conventional approach, DoE provides better flexibility of the process studies and a useful guideline for the membrane process operation for pX separation. (c) 2009 Society of Chemical Industry