TOC interpretation of lithofacies-based categorical regression model: A case study of the Yanchang formation shale in the Ordos basin, NW China

In this paper, taking the shale of Chang 7-Chang 9 oil formation in Yanchang Formation southeastern Ordos Basin as an example, through study heterogeneity characteristics, starting from preprocessing supervision data set, a logging interpretation method total organic carbon content (TOC) on lithofacies-based Categorical regression model (LBCRM) is proposed. It show that: 1) Based core observation, and Differences sedimentation structure, five lithofacies developed Formation: facies, siltstone/ultrafine sandstone tuff argillaceous facies with silty lamina lamina. 2) The strong makes it difficult to accurately explain TOC distribution intervals application model-based methods. LBCRM based understanding can effectively reduce influence factors other than prediction accuracy by studying characteristics constructing for each category. At same time, degree unbalanced reduced, so that mining algorithm achieves better effect. 3) interpretability ensures wellsite classification lithofacies. Compared traditional homogeneous model, performance has been greatly improved, segment more accurate. 4) understand reasons deviation method. After being combined latter, make provide useful information.