Pore Structure Characteristics And Classification Of Tight Sands From Shahejie Formation In Dongying Sag

The tight sands in the Bohai Bay Basin are high-quality reservoirs with abundant reserves and a high potential for exploration and development.However,they are characterized by extremely low permeability,complex pore structure,strong heterogeneity,micro-nano pore system,and low natural production,which require multi-stage hydraulic fracturing to achieve economic rates of production.Thus,the pore structure must be accurately and comprehensively investigated to reduce the exploration risk,evaluate the development potential effectively,and adjust the development schema.To this end,thin-section analysis,scanning electron microscopy,X-ray diffraction analysis,petrophysical properties analysis,porosity-permeability analysis,pressure-controlled porosimetry,rate-controlled porosimetry,and nuclear magnetic resonance were carried out on the tight sands of Es4~s and Es3~z in Dongying Sag to describe the petrological characteristics,diagenesis-evolution stages,petrophysical properties,and pore structure characteristics.On this basis,an improved method was developed and the dual pore structure model was improved to reveal the full-scale pore systems of tight sands.The classification method based on machine learning was also involved in the evaluation of the tight sands.An analysis of the reservoir petrological characteristics showed that the tight sandstone reservoir is mainly debris-arkosic sandstone with low compositional maturity and high clay matrix.The grain size is mainly fine and very fine,and the structural maturity is medium.Fine grain size is necessary to form a tight reservoir.The diagenesis analysis on the studied tight sands indicated that the existence of the phase A of the middle diagenetic stage and three main diageneses,i.e.,compaction,cementation,and dissolution.Influenced by the differential diagenesis process,a poor/non-existent reservoir and a relatively high-quality reservoir can be formed at the top/end and middle parts of sand bodies,respectively.The study on the reservoir petrophysical properties showed that the tight sands are mainly medium-ultra-low in porosity and ultra-low in permeability with an exponential relationship between porosities and permeabilities.Using thin-section and SEM techniques,primary intergranular pores,intergranular dissolution pores,intragranular dissolution pores,moldic pores,intercrystalline pores,and matrix pores can be observed in the studied tight sands.A close examination of the reservoir pore structure showed that the tight sands are characterized by nanometer sized throats,a large pore-to-throat ratio,high irreducible fluid saturation,and low movable fluid saturation,which has resulted in the low permeability of the reservoirs.The comparison between pore structure parameters and petrophysical properties revealed that the reservoirs are mainly controlled by the size of pores and throats.An improved full-scale pore structure characterization method,i.e.,the combination of pressure-controlled porosimetry,rate-controlled porosimetry,and nuclear magnetic resonance,is put forward to solve the problem that single reservoir pore structure characterization method cannot accurately obtain the full-scale pore size distribution of tight sandstone.The improved full-scale pore structure characterization method can obtain full-scale radius distribution curves of pore bodies and throats in a wide range of several nanometers to hundreds of microns,so as to achieve a more precise and comprehensive description of pore throat system of tight sands.Results showed that the micro pore bodies and nano throats are mainly found in the tight sandstone reservoir.The radius of pore bodies and throats are mainly in the range of 0.6-30?m and 0.03-1?m,respectively.The throat volume is larger than the pore body volume in the studied tight sandstone reservoir,which indicates that there are many nanoscale pores.It is difficult to distinguish the pore bodies and throats of these nanoscale pores.The fractal analysis of pore structure showed that two sections can be found in the pore structure fractal curves of most tight sand samples.Combined with the thin-section analysis and scanning electron microscopy observation,this paper considers that there are two types of pore structures with great differences in sizes,shapes,connections,and spatial distribution of pore bodies and throats,which can be expressed by the network-treelike dual pore model.The network structure comprises interconnected pore bodies and throats of varying sizes,and is characterized by large pore bodies and narrow throats.The treelike structure is characterized by no obvious difference between pore bodies and throats.The primary intergranular pores,intergranular dissolution pores,and moldic pores in tight sands have a network structure,while the intercrystalline pores and matrix pores have a treelike structure.The intragranular dissolution pores can have different structures under different conditions.In the samples from relatively high-quality reservoir with strong dissolution and poor cementation,throats are wide,greater than the radii of most intragranular dissolution pores.The intragranular dissolution pores can be filled by fluids in order of pore size after intergranular throats with connected pore bodies have been filled,which describe a typical tree-like structure.On the contrary,in the samples from relatively poor quality,throats are narrow,smaller than the radii of most intragranular dissolution pores.Most of the intragranular dissolution pores can be saturated directly after fluids enter intergranular throats,where intragranular dissolution pores exist as the pore part of network pores.Treelike pores are the main contributors to the storage capability of the studied tight sands with an average volume fraction of 62.8%,while the permeabilities are mainly contributed by network pores with an average permeability contributions of 93.35%.The pore structure classification of tight sands was carried out using a machine learning method.Classifications based on core and logging data were developed with an unsupervised learning method and a supervised learning method,respectively.Unsupervised learning results showed that five pore structure types can be divided in the tight sands.Supervised learning methods with non-linear models play better than linear methods in solving the nonlinear classification problem of pore structure logging evaluation of tight sands.Cross-validation showed that the accuracy of the ensemble learning method with four supervised learning classifiers was more than 90%and a good agreement was found between the prediction results of pore structure classification and oil productions.