3D Acoustic impedance modeling using turning bands simulation method in an oil field in SW of Iran

Petrophysical reservoir properties are usually estimated from elastic properties such as acoustic impedance (AI) using petro-elastic models. AI data are only available at sparse well locations, especially in early exploration stages of oil fields and therefore to quantify the spatial reservoir properties, it is necessary to estimate the AI for the rest of the reservoir area. Alongside the deterministic seismic inversion methods, multi-realization geostatistical simulation approaches can be used for the parameter estimations and uncertainty quantification. Geostatistical simulation methods allow the production of two- or three dimensional models of reservoir properties using existing well log data. Conventionally, sequential Gaussian simulation (SGS) method is used because of its simplicity. However, its accuracy is not always guaranteed. Nowadays, turning bands simulation method (TBSim) has received much attention because of its capability to reproduce the statistical properties of the original data. The main principle behind the TBSim is simplifying a multi-dimensional geostatistical simulation problem into a set of fast 1D simulation problems. The simulations are performed along the uniformly distributed lines spanning a unit sphere using sinusoid functions.  In this study, we are going to generate 3D AI models in an oil field in SW of Iran via the TBSim. We will also compare the results with the AI data computed from seismic inversion. Geostatistical modeling has less computational complexity than seismic inversion. Although AI modeling cannot be used as a definite alternative to seismic inversion, it can be used as a primary method for estimating AI, especially in areas without seismic data. It also has the capability to be involved with some seismic inversion methods known as stochastic seismic inversion methods. The dataset used in this study includes AI logs of seven wells, one of which (known as test well) has been excluded from calculations to check the accuracy of the results. The study reservoir zone includes the Ahwaz sandstone member in the upper part of Asmari formation and also the lower carbonates. The lower carbonates of Asmari formation are separated by an unconformity with Jahrum formation. After constructing the structural model and upscaling the AI logs in it, 50 realizations of AI are generated in the gridded model. The results in the test well indicate a high correlation between the modeled values and real AI data as well as the model obtained by seismic inversion. The maximum correlation of AI values of different realizations with real values in the test well equals to 78.8%. The correlation coefficient achieves to 82.9 % for mean of realizations and is 72.8% for seismic inversion data. The cube of mean of realizations is also in good agreement with the seismic inversion cube and reproduces the dominant trend of vertical and lateral AI variations. Comparison of results histogram with real data as well as the seismic inversion data reveals the capability of geostatistical AI modeling in reproducing the statistical properties of original data. In addition, the results of the uncertainty analysis of produced models also confirm the reliability of these models, especially in the test well. Therefore, we would recommend the TBSim as a powerful method for AI modeling during reservoir characterization.