The preparative separation of enantiomers of 1-1'-bi-2-naphthol on Pirkle covalent D-phenylglycine columns using hexane/isopropanol as the mobile phase was performed under a range of experimental conditions of flow rate, sample size, and mobile-phase composition. The system performance was evaluated based on the production rate with or without solvent consumption. Factorial design experiments using a spherical central composite design, with three variables (at five levels each) and the related response functions, were conducted to study the effect of the individual variables on the response functions. The regression models of the response functions were then established by second-order polynomials consisting of linear, quadratic, and interaction terms. The hybrid objective function (tradeoff between the production rate and solvent consumption) together with some constraints (e.g., constraints on purity, recovery yield, amount injected, and flow rate) were considered. The optimum level of these variables for obtaining maximum magnitude of the hybrid objective function were found and further verified experimentally.