Accounting for Seismic scale in Stochastic Reservoir Simulations Constrained by Seismic Data

Abstract Integrating seismic data in 3D reservoir property simulations leads to more accurate reservoir models. But the difference of scale between both data should be taken into account. For that, seismic data are usually considered as 2D information, hence related to the vertically averaged reservoir property; this implies that the seismic scale (or resolution) is identified to the reservoir thickness, i.e. to some user-defined size independent of the actual seismic volume support. In this context, we propose a methodology which takes into account a more realistic 3D seismic volume support. This methodology calls first for the determination of the 3D volume support of the seismic data by looking at seismic variograms or at the seismic signal. Then, a two-step simulation methodology is applied. In Step 1, the reservoir property cube at seismic scale (i.e., on the seismic volume support) is estimated from the seismic data. In Step 2, the resulting estimated cube is used for constraining the Sequential Gaussian Simulation (SGS) of the reservoir property (at point scale) either through cokriging of data at different scales or through non stationary kriging. An application of this methodology is presented: porosities are simulated at one reservoir level, constrained by a 3D impedance cube after inversion. Results of both Step 2 alternatives are compared in terms of input reproduction, easiness of implementation, computing time, impact of the seismic data,…