Aplicação de planejamento de experimentos em diferentes etapas do refino de petroleo

Abstract

In this work, experimental design technique was applied to different optimization studies, which were carried out in the Paulínia Refinery (Replan). The work covers applications of the technique in laboratorial scale, in computational tool, and the use of the technique in different industrial plants.The laboratory studies consisted in the evaluation of the dependence of the metropolitan diesel oil lubricity in relation to its composition and in the development of a new methodology to determine the flash point of light fractions of petroleum.Through factorial designs applied to a delayed coke simulator (named DC-SIM), simple polynomial equations were generated to predict the products yields of the unit in the reaction step. Using these equations, contour plots were obtained, which helped the analysis of the process response due to changes in the operating conditions. Two applications were done in industrial plants. First, using two level factorial designs, new models were adjusted to be used in the advanced control of a naphtha fractionation tower. The identification tests were carried out giving step disturbance in the manipulated variables. The experimental design methodology allowed a multivariable analysis of the system. Finally, experiments were done according to star configuration factorial designs in order to maximize the recovery of LPG molecules in the gases fractionating units of two delayed coke plants of Replan.Since Replan is the biggest oil refinery in Brazil, responsible for 20% of the national refining capacity, the use of factorial designs had a fundamental importance to allow the running of some tests in the plant. It reduces significantly the number of experiments, when it is compared to univariate methodology. Therefore, the impact of the tests in the production was minimized. Furthermore, the results provided very expressive financial earns.