Abstract

Acquiring the structures of porous media is very important when predicting flow properties in porous media. However, direct measurements of 3D microstructures of porous media with the resolution of microns or even nanometers are difficult to achieve due to the expensive cost of high-precision equipment. Therefore, as a typical stochastic simulation method, multiple-point statistics (MPS) was used to perform reconstruction based on real 3D volume data of porous media scanned by micro-CT. Because the ensemble of patterns extracted from a training image (TI) cannot be embedded into a linear space, the traditional MPS methods using linear dimensionality reduction, including filter-based simulation and distance-based pattern simulation, are not suitable to deal with the nonlinear situation. A new MPS method using isometric mapping, which is a method of nonlinear dimensionality reduction, to achieve nonlinear dimensionality reduction is proposed to decrease redundant data of TIs so that the subsequent simulation can be faster and more accurate for the reconstruction of porous media. Entropy theory is introduced to select a proper size of data template to balance the CPU cost and reconstructed quality. The comparisons between the reconstructed images and the target image show that the structural characteristics of reconstructed porous media using our method are similar to those of real volume data. This method also has shown advantages in reconstruction quality over typical methods using linear dimensionality reduction.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call