Response surface modelling and optimization in pervaporation

Both the conventional method of experimentation, in which one of factors is varied maintaining the other factors fixed at constant levels and the statistically designed experimental method, in which all factors are varied simultaneously are carried out for organic removal from water by pervaporation. Binary acetonitrile–water mixtures are considered. The effects of the operating parameters on the pervaporation performance of the membrane system have been investigated. The overall mass transfer coefficients have been determined for different conditions of feed temperature and initial organic concentration. In addition, the activation energy associated to the permeation process has been determined and discussed for each feed organic mixture. Statistical experimental design and response surface methodology, RSM, have been applied to optimize the operational conditions of pervaporation process in order to maximize the output responses, which are permeate flux ratio and concentration of organic in permeate. The input variables employed for experimental design were the feed temperature, initial concentration of organic in feed and operational downstream pressure. Based on the design of experiment the quadratic response surface models have been developed to link the output responses with the input variables via mathematical relationships. The constructed response models have been tested using the analysis of variance and the canonical analysis. The obtained optimal point by means of Monte Carlo simulation method and desirability function corresponds to a feed temperature of 57.69 °C, a feed acetonitrile concentration of 6.96 wt% and a downstream pressure of 28.95 kPa. The maximal values of the permeate flux ratio and the concentration of organic in permeate obtained under optimal process conditions have been confirmed experimentally.