Study On Channel Sand Body Connectivity Of Pu ? Formation In The Western Area Of Central Saertu Oilfield

The Pu I Formation in central Saertu Oilfield mainly belongs to fluvial and delta facies.Due to the incision and superposition of different channels,fluvial reservoirs often appear to be large-scale composite sand body with high heterogeneity.After decades of water flooding processes,Saertu Oilfield has already entered the extra-high water cut development stage,resulting in increasing difficulty for exploitation of remaining oil.In consequence,it is important to focus on subsurface reservoir architecture and sand body connectivity analysis.Based on subdivision and comparison of Pu I 2-3 thin layers in the western area of central Saertu Oilfield,the genetic types,identification marks,spatial distribution and contact modes of single channel sand body were described in detail.Furthermore,possible connection between two adjacent sand bodies and within single sand body were analyzed comprehensively,meanwhile,the key contributing factor was also summarized.Through these studies,channel sand body connectivity assessment system could be set up,and it is feasible to estimate practical connectivity by production performance,thus extract known samples from drilling or logging data of work area.In order to evaluate sand body connectivity automatically,support vector machine was applied to build quantitative models.If this learning algorithm is effective,it could be used to forecast reservoir connectivity of unknown regions,and would contribute to improving oil recovery.It was acknowledged that composite channel sand body could be divided into several single sand bodies.Identification marks for periodization included muddy interlayer,calcareous interlayer and petrophysical interlayer.In lateral dimension,composite sand body could be demarcated by abandoned channel,overbank deposits,size differences and some other signs.There existed four compound modes of subsurface sand bodies,including superposed type,cutting-overlaying type,isolated type and juxtaposed type,and the former two were more common than others in study area.Additionally,it has been found out that transverse connectivity between adjacent sand bodies was mainly affected by their genetic types and combinations,longitudinal connectivity should be considered with the impact of impervious interlayer between sedimentation units,and the seepage capability within single sand body was related to reservoir heterogeneity and structural features.On the basis of above findings,seven evaluation parameters of channel sand body connectivity were chosen,which included net-to-gross ratio,porosity and permeability of sand body,density and frequency of interlayer and so forth.181 learning samples were taken to train SVM models,and their predictive effects were verified with other 172 testing samples.The experimental results showed that prediction accuracy of SVM models could reach 90.7%,which indicated that support vector machine is a valid method in evaluation of channel sand body connectivity.