Multiple Regression Model for Determining and Predicting the Viscosity of Crude Oils Mixture

Abstract

The article presents development stages of a reliable multiple regression model for determining and predicting the oils mixture viscosity as a multifactor parameter. On the data of the laboratory experiment, a correlation and regression analysis was performed to select significant factors in the model. A fractional factorial experiment was carried out. A matrix of regression coefficients and a multiple linear regression equation were obtained. Estimation of the model significance has shown that the equation obtained describes empirical data with a high degree of reliability. The conducted studies showed that the known dependencies adequately describe the viscosity of crude oils mixture only when the content of a high-sulfur component is less than 10% or more than 90%. On a wider range of concentrations (20-80%), the viscosity of the mixture becomes a multifactor parameter and is more accurately described by the regression equation. The obtained dependence has a wide field of application in the practice of operating pipelines transporting compounded oil.