Reconstruction Of 3-D Digital Cores By Multiple-point Geostatistics Method Based On 2-D Images

With the increasing demand for energy,unconventional reservoirs,such as tight sand and shale,have been attracted more attention.The core from unconventional reservoir is tight and heterogeneous,which leads to complex rock physics properties.The uncertainty of rock physics has great impact on the accuracy of reservoir evaluation.The process of drainage and imbibition is hard to perform because of low porosity and permeability of tight sand.It is insufficient to reveal the relationship between rock physics properties and micro-structure quantitatively only by traditional rock physics experiments.Numerical simulation of rock physics based on 3-D digital cores has been an important method to study rock physics.The methods used to building 3-D digital cores for homogeneous siliciclastic cannot be directly applied to unconventional reservoir because of its heterogeneity.The X-ray Computed Tomography(CT)cannot recognize entire pore space although the space resolution has reached micron.So constructing 3-D digital cores that honor true microstructure from 2-D image is critical to study the effect of micro-factor on rock physics properties.Firstly,3-D digital cores are reconstructed by multiple-point geostatistics(MPS)based on single 2-D image.The reconstructed 3-D digital cores are quantitatively compared with the real cores by autocorrelation function,local porosity distribution function and average percolation probability function.The extracting method of hard data,the structure of searching tree and the algorithm of coupling condition probability distribution function of traditional MPS are revised to reconstruct 3-D digital core from three orthogonal 2-D images.3-D digital cores of Fontainebleau sandstone and Kaolinite filling are reconstructed by the improved MPS.The accuracy of reconstructed 3-D digital cores is evaluated by local porosity theory.Finally,we proposed a new method to reconstruct 3-D digital cores using MPS basedon medial axis of pore space.The medial axis is generated by MPS randomly and the pore space is reconstructed by the four parameter method.The codes of algorithm described above are written in MATLAB.The 3-D digital cores derived from single 2-D image have similar porosity and homogeneity with the real cores.But the connectivity of pore space is anisotropic,which is different with the true structure.The pore connectivity of the reconstructed 3-D digital cores based on three orthogonal 2-D images is isotropic and agree well with the true pore space in all three directions because the probability distributions from three directions are included as a constraint condition in MPS reconstruction.The new algorithm can be applied to reconstruct3-D digital cores based on scanning electron microscope(SEM)images to present pore structure with the size smaller than micron.Reconstructed 3-D digital cores based on medial axis of pore space have similar pore size distribution,homogeneity and connectivity with the real cores than that reconstructed from three orthogonal 2-D images by MPS.It provides a new method to reconstruct 3-D digital cores with variable pore structure.