Journal of Southwest Petroleum University(Science & Technology Edition) ›› 2026, Vol. 48 ›› Issue (5): 51-66.DOI: 10.11885/j.issn.1674-5086.2024.08.28.02

• GEOLOGY EXPLORATION • Previous Articles     Next Articles

Prediction of Favorable Reservoirs in Deep-water and Deep-layered Low-permeability Sandstones Based on Grain Size Distribution Patterns

ZHANG Chong, CHEN Ruxian, YE Qing, ZHOU Wei, LI Hua   

  1. CNOOC China Limited, Hainan Branch, Haikou, Hainan 570312, China
  • Received:2024-08-28 Published:2026-11-03

Abstract: Due to the poor quality of deep water seismic data in Baodao B Area in Qiongdongnan Basin, it is difficult to predict favorable lithology. Based on the experimental data of thin section, particle size and minerals of the third member of Lingshui Formation, the grain size distribution characteristics of low permeability sandstone are analyzed. Secondly, the lithology probability sensitive factor curve based on conventional logging data is constructed by using Fisher discriminant analysis method. On this basis, a frequency-division intelligent reservoir inversion method based on lithology probability sensitive factor is proposed to predict favorable lithology distribution. The spatial distribution of favorable lithology is revealed. The results are as follow: 1) The reservoir rocks are characteristic of wide particle size distribution, poor sorting and positive skewness. When the particle size is less than 3.3 ϕ, the permeability is generally greater than 1 mD. According to the grain size classification criteria, the reservoir lithology can be further divided into relatively coarse-grained sandstone (grain size less than 3.3 ϕ), relatively fine-grained sandstone (grain size 3.3~4.0 ϕ) and siltstone-mudstone (grain size greater than 4.0 ϕ). 2) The lithology probability sensitivity factor curve based on FDA discriminating principle is constructed, which can accurately identify the three types of lithology, and the accuracy rate is 27% higher than that of a single logging cross plot. 3) The frequency-division intelligent inversion method of lithology prediction by reconstructing the lithology probability sensitivity factor curve greatly improves the lithology prediction accuracy, and the prediction results show that the relatively coarse lithology is mainly distributed in the northern part of the sedimentary body, which is a potential favorable reservoir distribution area. The research results can provide geological basis for sweet spot evaluation of deep water and low permeability gas reservoirs.

Key words: low-permeability reservoir, Fisher discriminant, lithology probability sensitivity factor, BP neural network, frequency division intelligent inversion, deep-water and deep-layered, Baodao Sag

CLC Number: