Study On Fluid Identification Method Of Complex Carbonate Reservoir In Middle And Lower Assemblage Of M Formation In JX Area

Carbonate reservoir plays an important role in the history of oil and gas exploration and development.But as a result of carbonate reservoir with strong heterogeneity,complex reservoir space types and the characteristics of low permeability reservoirs,the carbonate reservoir in physical property,oiliness prediction as well as fluid property identification,etc,becomes more difficult,so reasonable and effective to carry out the reservoir logging evaluation becomes more important,also to the exploration of carbonate reservoir development has very important theoretical significance and practical application value.This paper M middle group in JX area carbonate reservoir as an example,the combination with lithologic data and analysis data of physical properties,testing data and conventional log data,analysis of the study area reservoir lithology,physical property,oiliness,electrical,and reservoir space types and so on characteristics,has been clear about the main control factors of reservoir,lay a foundation for reservoir fluid identification.On the basis of the four properties analysis,combined with the physical property data of core analysis and conventional logging data,the modeling work of reservoir porosity,permeability and water saturation is carried out respectively,and the multi-mineral optimization model is applied to the calculation of reservoir porosity in the study area,which obtains good results.The machine learning method is introduced into the permeability modeling.According to the permeability differences of different pore types,the reservoir permeability models based on intergranular pores,fractures and dissolved pores are established respectively,and the reservoir permeability is calculated with the combination of neural network.The application effect is good.Due to the heterogeneity of carbonate reservoir,using the traditional reservoir water saturation in the archie formula has become not applicable,will have been used in igneous rocks and achieved good results of water saturation model based on "harbor effect" is applied to the research area of carbonate water saturation model,the calculation result is better.Intergranular pores,fractures and dissolved pores are developed in the carbonate reservoirs of the lower and middle assemblages of M formation in JX area,which are characterized by large distribution variation,irregular morphology,uneven development and strong heterogeneity,and the pore types are relatively complex.In the process of exploration,it was found that some Wells in the study area had high resistance to water,and the gas-water relationship was complex and the fluid identification was difficult.Respectively in this paper by using conventional logging data,array acoustic logging data and electric imaging logging data,and uses the Fisher discriminant,the Bayes discriminant,the SVM discriminant,the KNN discriminant and the Random forest discrimination five machine learning methods,to identify the laminar liquid storage,not only can be a very good discriminant atmosphere,gas,water,gas water namely,gas water,poor gas and dry layers 6 fluid types,and by conventional logging data with the combination of electric imaging logging data method can effectively identify the high resistance layer.