Evaluation Techniques for Shale Oil Lithology and Mineral Composition Based on Principal Component Analysis Optimized Clustering Algorithm

Abstract

Shale oil reservoirs are characterized by complex lithology, complex mineral composition and strong heterogeneity. This causes great difficulty in lithologic evaluation. In this paper, a method of lithology identification is proposed by means of intersection plot method and machine learning method, and lithology evaluation is carried out by combining the calculation of mineral content with a multi-mineral optimization model. The logging response characteristics of five lithologies are analyzed by using the logging curves selected by principal component analysis (PCA) discriminant analysis. In lithology identification, the system clustering algorithm is selected to identify shale oil reservoir lithology through layer-by-layer subdivision of sample lithology classification. Logging data has high vertical resolution and good continuity, and mineral prediction using logging data can ensure high accuracy. In this paper, the method of calculating mineral content by using multi-mineral optimization model has achieved good results in practice.