--> Study on Flow Unit of Turbidite Fan Low Permeability Reservoir Based on Self-Organizing Neural Network Algorithm

2019 AAPG Annual Convention and Exhibition:

Datapages, Inc.Print this page

Study on Flow Unit of Turbidite Fan Low Permeability Reservoir Based on Self-Organizing Neural Network Algorithm

Abstract

Daluhu Area is located in the northwestern part of Boxing Depression in Dongying Sag, Jiyang Depression, Bohai Bay Basin, China. The Middle Es3 Member of the Paleogene in Daluhu area is a typical sedimentation of turbidite fans. The reservoir architecture is complex and the sand and mud combination is diverse. At present, the study on reservoir architecture and genesis mechanism of turbidite channel reservoir has been carried out. The research of reservoir architecture is of great significance for the study of flow unit. Therefore, the distribution of flow unit based on reservoir architecture has become the focus of current research. Based on available reservoir architecture division,this paper quantitatively evaluated the flow unit by applying the self-organizing neural network in the supervised mode. The performance of the model was tested by training samples. Finally, the distribution of flow unit was analyzed based on the study of reservoir architecture. The results are of great significance for oilfield development and residual oil prediction. The results show that: (1) According to the genetic type and sedimentary model of the turbidite channel, the reservoir in the study area can be further divided into three types of single channel. The relationship between seepage barriers and architecture surfaces was summarized from the levels of single channel, composite channel and channel system. (2) The algorithm of self-organizing neural network can output the optimal prediction model by means of neuron competition learning and mutual supervision. The accuracy rate of the training samples reached 89.34%. The prediction results of the flow unit are basically consistent with the qualitative division results and the initial productivity characteristics of the wells. The algorithm provides a new means for the study of flow units. (3) The reservoir architecture has a great influence on the flow unit distribution. Vertically, the flow unit in different types of single channels may vary greatly, causing the difference of flow unit distribution; Laterally, influenced by the sedimentary evolution stages of channels systems, lateral division of flow unit is different; Horizontally, due to the migration and vertical accretion of the single channels, the distribution of flow unit in the composite channel is significantly different.