Identification of interlayer and connectivity analysis based on machine learning and production data: A case study from M oilfield

Interlayer is an important factor affecting the distribution of remaining oil. Accurate identification of interlayer distribution is of great significance in guiding oilfield production and development. However, the traditional method of identifying interlayers has some limitations: (1) Due to the existence of overlaps in the cross plot for different categories of interlayers, it is difficult to establish a determined model to classify the type of interlayer; (2) Traditional identification methods only use two or three logging curves to identify the types of interlayers, making it difficult to fully utilize the information of the logging curves, the recognition accuracy will be greatly reduced; (3) For a large number of complex logging data, interlayer identification is time-consuming and labor-intensive. Based on the existing well area data such as logging data and core data, this paper uses machine learning method to quantitatively identify the interlayers in the single well layer of CⅢ sandstone group in the M oilfield. Through the comparison of various classifiers, it is found that the decision tree method has the best applicability and the highest accuracy in the study area. Based on single well identification of interlayers, the continuity of well interval interlayers in the study area is analyzed according to the horizontal well. Finally, the influence of the continuity of interlayers on the distribution of remaining oil is verified by the spatial distribution characteristics of interlayers combined with the production situation of the M oilfield.