Digital core analysis based on micro-CT computed tomography provide a new approach in core samples’ permeability numerical prediction. Firstly, Minkowski functionals is applied along with the K-means cluster analysis to separate the sandstone core sample gotten from the micro-CT computed tomography into distinct classes, well-sorted fine grain, well-sorted large grain and transition area in order to get the elaborate layer system. Secondly, capillary drainage transform is performed to simulate the distribution of wetting and non-wetting phases under different drainage pressures. Lastly, LBM method is operated based on the core layer system to model permeability of each layers and upscaling is applied based on the simulated results with 3 different ways: directly applying the average method on the prediction values, finding the correlation function of permeability and porosity with and without thin layers in order to get the upscaling values of permeability and correlation function of permeability and porosity.