The Study On Petrophysical Facies Characterization Method Of Bioclastic Limestone Reservoir In West Qurna Oilfield,Iraq

Controlled by both sedimentation and diagenesis,the bioclastic limestone reservoir of Mishrif Formation in West Qurna Oilfield shows the characteristics of multiple complex mineral texture and composition,complex pore structure and reservoir genetic types.This paper takes artificial intelligence methods as the key data-driven technology to make quantitative classification and intelligent identification of reservoir petrophysical facies to make high-precision prediction of reservoir parameters.The scientific problems such as the quantitative characterization of geological characteristics difference,reservoir quality difference,logging response difference,and seismic wave impedance difference and permeability prediction under three-dimensional space are solved in this article.Based on those study,petrophysical facies characterization methods and procedures suitable for bioclastic limestone reservoir are formed,reservoir petrophysical facies distribution model of bioclastic limestone reservoir are finally established,and distribution law of reservoir petrophysical facies are clarified,which provide basis for efficient development of oilfield.Because reservoir data are various in types,resolution and volume,this paper proposes the sliding window method for the automatic core restoration and the intelligent seismic well tie to realize the matching and high unification of information from different sources and scales in bioclastic rock reservoirs,which has laid a foundation for the establishment of reservoir data frame.Aiming at the quantitative classification of rock petrophysical facies in bioclastic limestone controlled by sedimentation and diagenesis,thin section data are used to quantitatively characterize the lithology and diagenesis by computer vision image segmentation and parameter extraction technology.Finally,those thin section data are divided into 10 types of geological petrophysical facies(GPF)with clear lithology or diagenesis differences.For the quantitative characterization of the relationship between discrete data and continuous data,a hierarchical clustering algorithm based on grid and density is developed.As a result,four types of reservoir physical facies(RPF)with obvious difference in reservoir quality and porositypermeability relationship,eight types of logging petrophysical facies(LPF)with obvious difference in logging response and three types of seismic petrophysical facies(SPF)with obvious difference in wave impedance are excavated.Aiming at the problem of poor generalization ability of the model caused by uneven distribution between label data and prediction data in conventional machine learning algorithm,a differential nearest neighbor algorithm is developed to identify the logging petrophysical facies.The result show that the accuracy rate reaches 95.5%,and the prediction results strictly follow the geological interface,thus obviously improving the generalization ability of the model.Based on the logging petrophysical facies prediction results,the application of reservoir petrophysical facies technology to the whole well section reduced the average absolute error of permeability prediction from 26.03 m D to 19.07 m D,significantly improving the accuracy of permeability prediction.The hierarchical calibration methods including the thin section-core calibration,core-logging calibration and logging-seismic calibration are adopted to characterize the microfacies of the sedimentary slope of the bioclastic limestone reservoir.The based on graph theory rolling maximum inscribed circle algorithm is first proposed for the quantitative characterization of the tidal channel network morphology and channel parameters.Based on the grid and density overlapping hierarchical clustering algorithm,quantitative characterization of reservoir quality difference between sedimentary microfacies is carried out,and the geological rules that the lithologic distribution are controlled by deposition energy of sedimentary environment and the diagenesis type are controlled by the tidal horizon are excavated.Under the guidance of this rule,combined with the reservoir petrophysical facies distribution,the reservoir petrophysical facies model in the study area is established.Under the guidance of the reservoir petrophysical facies model,spatial coupling of sedimentation and diagenesis by applying wave impedance inversion,Kriging inter-well collaborative simulation,quantitative seismic sedimentology for tidal channel characterization,interfacial constrained deterministic modeling technique and other technologies,so as to realize identification of reservoir petrophysical facies and highprecision prediction of permeability in three-dimensional space.