Aplicação de planejamento de experimentos na etapa de clarificação de óleo de soja em planta piloto

Abstract

The process of obtaining soybean oil consists in a series of physical and chemical steps aiming to removing impurities, one of the steps is the clarification process that consists in the adsorption of pigments and other components present in the oil with the use of a mineral adsorbent agent. The clarification step involves a large number of factors that must be optimized because they directly impact on numerous response parameters of the procedure. In order to this optimization the present study proposed the analyse the soybean oil clarification in a pilot plant with the use of two different filter auxiliary in order to identify the significant effects of the factors on the responses variables and in order to obtain mathematical models that can determine optimal points for operation of the clarification process. A central composite planning with four process factors (temperature, adsorbent concentration, filter auxiliary concentration and time) and three responses (chlorophyll, Lovibond yellow and Lovibond red) was performed, which were considered important for the clarification step. Three experimental points were added to the experimental design to estimate the experimental error. The experiments were carried out for two different types of filtering auxiliaries, one mineral and the other cellulosic. The effects of each parameter adopted on each response variable for each filter auxiliary studied were quantified allowing the determination of the significance of each parameter. The methodology of the desirability function was applied to determine the optimal parameters of operation in the clarification stage by weighing each response variable in individual occurrence desires and then calculating the overall desirability of the responses. The non- parametric Mann-Whitney test was applied to verify if there was any difference in the results obtained between the two filter auxiliaries. By the obtained results it was possible to determine graphical regions where the statistically significant parameters have greater effect on the studied variable. The desirability function resulted in the optimum overall desirability equal to 1 for the results obtained with the mineral filter auxiliary using 2% adsorbent concentration, 0.5% filter auxiliary concentration, 65 minutes of process at 140 ° C and 0.90901 for the results obtained for the cellulosic filter auxiliary using 2.625% adsorbent concentration, 0.1% filter auxiliary concentration, 35 minutes of process at 90° C. The Mann-Whitney test between the results obtained for each filter auxiliary studied demostred that for the variable chlorophyll there is no significant difference between the two filter auxiliaries, but for the variables yellow and red Lovibond, the are significant difference. By the analysis of the results of the Mann-Whitney test, it was verified that the results obtained for the mineral filter auxiliary was more efficient for the reduction of the studied variables variable. The planning of experiments employed provided statistically significant and coherent answers, being this a great tool of decision making. For the reduction of the response variables studied it is observed by the results that the mineral filter aid is more efficient.