Quantitative Seismic Reservoir Characterization Based On Geostatistics And Rock Physics

Oil and gas reservoirs are the direct target layer for petroleum exploration and exploitation.Reservoir properties such as lithofluid facies,porosity,fluid saturation and shale volume have an important influence on the reserves evaluation and the determination of optimal well placement.However,due to the bandlimited characteristics of seismic data,it is very challenging to accurately predict the distribution of reservoirs or to quantitatively characterize the reservoir properties just by a single prestack or poststack seismic inversion.Therefore,we adopt the idea of the integration of multi-data,multi-discipline and multi-method,and combine geostatistics,rock physics and seismic inversion to establish a workflow for the quantitative seismic reservoir characterization.In the rock-physics modelling,we propose an extended Xu-White model for the complex pore structure reservoirs.By introducing a third pore structure,the analytical expression of the equivalent modulus of the multi-porous dry rock is derived based on the differential equivalent medium theory,then the fluid saturated rock modulus are calculated by the pathcy saturation model.The proposed model can provide more accurate total porosity estimation and shear wave prediction results,and can quantitatively predict the porosity and gas saturation from the seismic inversion results.Since the seismic wavelet and noise are important input parameters for seismic inversion,and the conventional seismic wavelet extraction method does not consider the coupling relationship between seismic wavelet and noise.We use the Bayesian network to describe the dependency of various parameters in the well-log constrained seismic wavelet extraction,and consider the calibration error and measurement error of well-log data to estimate the seismic wavelet and noise simultaneously and quantitatively.The estimated results leads to a better well-to-seismic tie result.Due to the advantages of high signal-to-noise ratio and low computational cost of poststack seismic data,the impedance inversion is still widely used in the industry for qualitative and quantitative reservoir characterization.But the impedance inversion cannot accurately distinguish the gas layer from the brine layer when their impedance overlaps.Hence,we propose a new strategy for the integrated prediction of favorable reservoir by using the poststack impedance inversion and the frequency-dependent fluid mobility attribute.The integrated method combines some advantages of both the seismic impedance inversion and the fluid mobility attribute,including high resolution,stratigraphic interpretation and accurate gas detection.For prestack seismic inversion,to overcome the low resolution issue of the conventional deterministic seismic inversion and the computational inefficiency issue of the traditional stochastic seismic inversion,we propose a high resolution prestack Bayesian sequential stochastic inversion method.It combines seismic data,well-log data and geostatistical information together under the Bayesian framework.This method is a contribution to both seismic inversion theory and geostatistics.The 1D and 2D theoretical model tests and the 3D real data application of the lithofluid quantitative characterization are performed.The results verify the advantatges of the proposed method in improving the accuracy of prestack seismic inversion and reducing the uncertainty.