Classification of Hard-to-Recover Hydrocarbon Reserves of Kazakhstan with the Use of Fuzzy Cluster-Analysis

摘要

This report is devoted to the classification of hard-to-recover oil reserves. The analysis of existing classifications has been carried out preliminary and the necessity of using a method that takes into account the whole range of characteristics allowing to classify oil and conditions of occurrence to a particular class has been shown. In this connection, we applied the method of fuzzy cluster analysis. The tasks of cluster-analysis have been widely used in economics, sociology, medicine, geology, oilfield practice and other industries, i.e. wherever there are sets of objects of an arbitrary nature, described in the form of vectors x= {x_1, x_2,...,x_N}, which must be automatically divided into groups of homogeneous objects according to the similarity within the homogeneous object (cluster) and the difference between these objects. A considerable amount of literature has accumulated in this direction. As noted in the literature, there are more than one hundred different clustering algorithms, among them hierarchical and non-hierarchical cluster-analyzes, fuzzy clustering. In order to classify hard-to-recover reserves, we performed clustering using the fuzzy cluster-analysis algorithm. For this purpose, data were collected on the viscosity, oil density and permeability of oil conditions from the oilfields of Kazakhstan. As a result, 4 classes were obtained, each of which characterizes the difficulty of extracting oil: the layer is permeable, highly viscous and very heavy oil; medium permeability layer, viscous and heavy oil; high-permeability reservoir, medium viscosity oil and medium-density oil; low-permeability reservoir, low viscosity oil, light oil.