PREDICTION OF GEL POINT FOR MULTI-COMPONENT BLEND CRUDE OIL USING SUPPORT VECTOR MACHINE ALGORITHM

摘要

A solid-like structure with a certain yield strength will be formed when the oil temperature goes below its gel point. The oil temperature should be kept above its gel point during the pipeline transportation. It is hard to predict the gel point for multicomponent blend crude oil, since the gel point of each single kind of oil will affect that of the blend oil in varying degrees, and the influence of single oil property and mixing proportion on the gel point of the multi-component blend crude oil is quite complicated with nonlinear characteristics. Considering the calculation of gel point for blend crude oil with a variety of components and mixing proportions as a multiple non-linear regression analysis, a prediction model for blend oil gel point is obtained by applying support vector machine algorithm, and the prediction result is reliable with a good accuracy.