Study On The Log Interpretation Method Of Low Contrast Oil Pays In Tight Sandstone Reservoir Of Longdong Area

The tight sandstone reservoir of Chang8 member in Longdong area of Ordos Basin has low porosity and permeability,strong heterogeneity,complex pore structure and oil-water relationship.The contribution of pore fluid to logging response is small,and the difference in electrical and physical properties between oil layer and water layer is not obvious,low contrast oil reservoirs are developed.It is difficult to qualitatively identify and quantitatively evalua te this kind of reservoirs,which seriously restricted the exploration process and development benefit of oil resources in this area.In this study,the petrophysical characteristics and the relationship between"four properties"of low contrast oil reservoir were first analyzed based on the reservoir geological characteristics,petrophysical experiment,logging response characteristics and digital rock technology.And the geneses of low contrast oil reservoir were studied from the micro and macro factors of the reservoir.On this basis,the fluid identification factors were constructed,and the double apparent Rw curve overlapping method,the partition cross plot method,the full hydrocarbon mud logging and well logging hybrid method and the P^1/2 normal distribution method were established.A comprehensive fluid identification method was proposed based on the coincidence rate of cross-plots and voting strategy,and the processing program of this method was realized by using the Forward.net software.Besides,the fluid identification methods of NMR and array acoustic logging were optimized,and the recognition effect of low contrast oil reservoir was explored.Then,the appropriate models of reservoir porosity,permeability and saturation were established and optimized.The variable Archie index model,dual porosity model,equivalent rock element model?EREM?and variable cementation index?m?model were established and their calculation results were compared in different pore structure regions.In addition,from the perspective of data mining,SVM classification model and support vector regression?SVR?model for fluid identification and reservoir parameter prediction respectively were constructed by using support vector machine?SVM?method.Finally,the proposed methods and models were applied in our research area and the effect is analyzed.It is found that the high irreducible water saturation caused of the development of micropores,coupled with high water salinity and complex rock wettability are the main micro factors leading to low contrast oil reservoirs.In addition,the regional difference of formation water salinity and hydrocarbon expulsion ability of source rock are the macro factors controlled the distribution of low contrast oil reservoirs.The proposed comprehensive fluid identification method not only considers the independent judgment of different methods,but also reflects the effective combination of various methods.The application effect is better than the single cross plot method,and the accuracy of conventional logging fluid identification is improved.Besides,the fluid identification method based on NMR and array acoustic logging reduced the influence of rock skeleton,and identification effect of complex oil-water layer is enhanced.The porosity calculation model based on multiple regression of logging curves,and the permeability calculation model based on NMR logging are more accurate.Comparing the calculation results of different saturation models with the experimental data of sealed coring,it is found that the calculation results of variable m model based on the capillary bundle model are in good agreement with the core analysis data,which proves that the saturation model based on the pore structure characteristics is effective.The fluid identification accuracy of SVM classification model is higher than that of conventiona l cross plot method and BP neural network classification model.The reservoir permeability and water saturation predicted by SVR regression model in the study area are more accurate than that calculated by conventional model established by experimental data,which shows that it is feasible and effective to carry out logging interpretation and evaluation of complex oil-water layer based on SVM method.The above research results can provide important reference and interpretation basis for oil exploration and development and old well re-examination in the study area.