致密砂岩储层复杂油水层成因分析与测井评价OA北大核心

The tight sandstone reservoir has strong heterogeneity and complex pore structure.Weak geo-logical signals increase the multiplicity of logging interpretation,and it is difficult to identify oil and wa-ter layers.Taking the sandstone reservoir of the 6th member of Yanchang Formation in Ningxian Area of Ordos Basin as the research object,the genesis of complex oil and water layers in the study area was analyzed by experiments on lithology,physical property,pore structure,fluid property and fluid occur-rence state of the cores.The BP neural network method is used to establish reservoir physical property and oiliness logging interpretation models,the fluid property identification model is established by cross-plot method and random forest method.The results show that the additional conductivity of rock clay minerals,the blockage of pores by carbonate cements and carbonaceous asphalt,the crude oil in the form of bound state and semi-bound state are the main factors for the complex oil-water relationship in the tight oil reservoirs of the 6th member of Yanchang Formation in the study area.The interpretation models of reservoir physical property and oiliness established by BP neural network method,which can improve the calculation accuracy of reservoir parameters.The intersection chart of porosity and oil satu-ration can effectively identify oil layers,oil-water layers and water layers(including oil-bearing water layers).The fluid property identification model based on random forest algorithm can accurately distin-guish oil layer,such as class Ⅰ oil-water layer(more than 2 t/d),class Ⅱ oil-water layer(less than 2 t/d)and water layer(including oil-bearing water layer).The results can improve the coincidence rate of log-ging interpretation greatly and provide technical support for the economic development and effective uti-lization of reservoirs in the study area.