Geostatistical modeling of porosity and evaluating the local and global distribution

Abstract Porosity is one of the main variables needed for reservoir characterization. For this volumetric variable, there are many methods to simulate spatial distribution. In article, porosity was analyzed and modeled in local global simulation, Sequential Gaussian simulation (SGS) Random Function (GRFS) were applied. Also, kriging used estimate at specific locations. The purpose work investigate compare geostatistical estimation a sandstone as real case study. First, data sets normalized by Normal Scores Transformation (NST) stratigraphic coordinate. model experimental variograms fitted vertical horizontal directions. methods, 10 realizations generated each method. Q-Q plots calculated, both quintiles (Target Distribution versus realization) came from normal distributions with following correlation coefficients: 0.93, 0.94 0.97 related GRFS, SGS Kriging, respectively. extracted showed that couldn’t reproduce validation, cross-validation evaluated three wells omitted. re-estimation considered located well logs through sections window where had better performance minimum error locally. Finally, cross-sectional models algorithm which simple tries produce smoother distribution, whereas conditional simulations (SGS GRFS) try represent more global-detailed sections.