Study On Gas Detection Methods For Granite Weathering Crust Reservoir In Qiongdongnan Basin

Bedrock reservoirs are an important type of reservoir.In recent years,bedrock reservoirs of granite weathering crust have been found in Bohai Sea and South China Sea.Granite weathering crust reservoirs have complex and diverse physical properties,and strong heterogeneity in vertical and horizontal directions.It is very difficult to characterize reservoirs and detect gas-bearing properties.This thesis focuses on the study of the characterization and gas bearing detection of the weathering crust reservoir of granite in an exploration area of Qiongdongnan Basin.The concrete results of the paper are summarized as follows:1.Identification and characterization of granite weathering crust reservoir in the study area.The impression method is used to restore ancient landforms in the study area,and the relationship between the time height of ancient landform and the time thickness of weathering crust was fitted.The fitting results are extended to the whole area to quickly complete the calibration of the weathered crust bottom interface in the study area;On this basis,the longitudinal zoning of granite weathering crust in the study area was carried out by combining with geological background,wall core,logging data and electrical imaging characteristics to find out high-quality weathering zones,and the longitudinal zoning of weathering crust,seismic response and other characteristics were studied.2.This thesis introduces the theoretical basis of several methods for gas bearing detection in reservoirs.The theoretical basis and realization process of AVO anomaly analysis(AVO inversion,elastic impedance inversion,pre-stack simultaneous inversion)are mainly expounded.The principle of recursive feature elimination method based on support vector machine(SVM-RFE)and the flow of fluid recognition method based on SVM-RFE are studied.The theoretical basis of dispersion analysis and impedance inversion of post-stack wave are introduced in detail.3.In this thesis,a variety of gas bearing detection methods are used to detect the gas bearing of weathering crust reservoirs,including:(1)The time-frequency analysis of the seismic data in the weathering crust section of the study area was carried out by using the generalized S transformation,and the low-frequency attenuation gradient property was obtained by fitting the low-frequency part of the seismic wave frequency band with the linear method for the detection of gas content.(2)The intercept P and gradient G attributes were extracted from the pre-stack CRP trace set for AVO attribute analysis,and the PG and other derived profiles were generated for gas-water identification based on the intercept P and gradient G.The profile distribution of such attributes are somewhat disorder and can not accurately predict the distribution of gas-bearing reservoirs.(3)Constrained by well logging and geological data,the longitudinal wave impedance,shear wave impedance and density are simultaneously obtained by using the prestack angle gather,and the gas-bearing property is predicted by calculating elastic sensitive parameters based on the analysis of reservoir sensitive parameters.The results obtained by prestack simultaneous inversion can effectively predict the reservoir fluid distribution in the research area.(4)The fluid identification method based on SVM-RFE uses well logging data to calculate a large number of elastic parameters and fluid factors to form a sample set,from which the optimal fluid identification combination is selected,and the SVM-RFE regression model is trained to predict the fluid parameters of the target region,and a good inversion result is obtained.4.It is considered that the dispersion analysis is simple and efficient,but the application effect is poor.The AVO attribute analysis is not effective in deep reservoir due to the influence of the strong heterogeneity of granite weathering crust reservoir.The elastic parameters such as mu-rho,lambda-rho and density obtained from pre-stack inversion can predict the distribution of gas-bearing reservoir more accurately.The fluid prediction method based on svm-rfe is trained with multiple fluid identification factors as sample sets,which can predict the weathering crust gas-bearing reservoirs more accurately and reliably.