Study On Quantitative Description Of Micro Residual Oil After Polymer Flooding Based On Digital Pore And Throat

With the rapid development of the national economy,petroleum is continuously extracted as an important energy source,and the supply of oil and gas resources is in short supply,which makes people realize the importance of using effective means to tap the remaining oil.After several years of mining,Daqing Oilfield was in a stage of high water cut,and the percentage of water content was more than 95%,but there are still a large number of unexplored reserves.In-depth study of the remaining oil,understanding the cause-s and mob ilization conditions of the remaining oil,formulating potential tapping countermeasures,and using the oil displacement means to convert the remaining oil hidden in the for-mation into recoverable movable oil,which is targeted for the develo-pment of high water content.It is important to guide significance for chemical flooding formula selection.Porous media was studied in this paper.The distribution frequency of pore throat and the coordination number measured by CT scan are measured by mercury intrusion test,and the shape and wettability of the throat are considered.The adaptive porosity method is used to fit the measured value.After the permeability,a digital pore network model is constructed.Considering the non-equal diameter characteristics of the pore throat space,a digital three dimensional network model of asymmetric corrugated tubular pore channels conforming to th e actual core pore structure is established,and the throat model is visualized;The relevant for mulas of seepage mechanics,and fluid mechanics are using in this process,The modelof oil saturated,water flooding and polymer flooding is established by considering the change of oil-water interface in the throat,and the saturated oil,thewater flooding and the Polymer flooding are simulated.The remaining oil images of different shapes are distinguished and labeled.Recognition,extraction and cutting of various remaining oil images are using by the Deep Lab V3+ deep learning network.The remaining oil images were classified by using the convolutional neural network Mobile Net V1 and Mobile Net V2 were identified using the improved Mobile Net V2 convolutional neural network;The accuracy,recall rate and accuracy rate of the original and improved Mobile Net convolution neural networks are compared,and the superiority of the improved Mobile Net convolution neural network in the task of remaining oil image recognition of the digital pore network model in this paper is proved.The effects of pore structure and the proportion of each component of polymer flooding on the flooding efficiency and residual oil type were analyzed.The microscopic residual oil occurrence state,residual oil content distribution and types of residual oil distribution in the pores after polymer flooding were predicted.Through our research,not only the visualization of the microscopic residual oil occurrence state in the pores of the reservoir after polymer flooding can be realized,but also the theoretical and data basis for the subsequent mining can be provided.The results have been indicated that: Based on the measured pore structure parameters,the three-dimensional network model of asymmetric corrugated tubular pore channel established by the adaptive porosity principle can effectively describe the complex pore structure through the comparative analysis of porosity and permeability.Based on the three-dimensional pore structure model,considering the adsorption and retention characteristics of polymer solution,the dynamic simulation of water flooding and polymer flooding process and the visualization of oil-water distribution in each flooding stage model are realized.The distribution law of remaining oil after model polymer flooding is basically consistent with that determined by CT scan after core physical simulation experiment,which is verified the correctness by model simulation process.