4D Feasibility and Calibration Using 3D Seismic Modeling of Reservoir Models

Abstract

Abstract Monitoring fluid motions in a producing reservoir is a key to understand reservoir behavior and anticipate drilling decisions and investments. In a deep-water environment such as Girassol (Angolan deep water, West Africa), the cost of a 4D seismic is less than the cost of a well, and 4D seismic is expected to provide the necessary monitoring information in terms of saturation and movement of fluids. Meanwhile, before deciding to repeat a 3D acquisition, the main question to answer through a feasibility study is: will the dynamic changes occurring in the reservoir be visible on seismic data? To achieve the 4D feasibility study, a reservoir model representing the state of the reservoir before production/injection, and another reservoir model after a certain time of production/injection were built. 3D Pre-stack synthetic seismic responses of the models were then computed in time and in depth using a fast algorithm accounting for fine-scale details of the reservoir model. Average amplitude maps extracted at key horizons, from the difference between 3D synthetic seismic responses of the two reservoir models showed the anticipated lateral extent of injection and production, and the impact in terms of amplitudes of the different injection and production mechanisms. These results of the 4D feasibility made us confident in the fact that the forecasted 3D seismic acquisition would provide important information about the reservoir dynamic characteristics. Since this work has been conducted, the new 3D seismic data that have been acquired. The identification and understanding of the differences between real and synthetic data helped us qualitatively and quantitatively to update the reservoir model and the petro-elastic model. Introduction The 4D seismic technology is used to update reservoir models and contributes to the field development plan: infill drillings, detection of undrained compartments, Stronen et al.1, gas or water injection monitoring, faults retention capacity assessment, Sonneland et al.2. The update of reservoir models using 4D seismic data is not direct, and the link between pressure and saturation changes and amplitudes, can be performed through seismic modeling using various reservoir simulation grids at different stages of production, Figure 1. However, creating pre-stack seismic cubes from 3D reservoir grids including millions of cells is not an easy task, Yuh et al.3, Lumley et al.4. 4D feasibilities are often done in 1D or 2D, with pressure and saturations values coming from PVT laws. This precludes the visualization of 3D seismic effects due to reservoir changes such as injected gas and water lateral extension or depleted areas. In this paper, a 3D pre-stack seismic modeling strategy well suited for 4D techniques is presented. Fast and efficient, it computes the seismic response according to reservoir simulation results. This allows to calibrate the 4D seismic data interpretation, analyzing the impacts on amplitudes, time shifts and spatial organizations of the pressure and saturation changes in a real 3D framework. The methodology consists of 5 steps: reservoir flow modeling, preparation of reservoir grids for seismic modeling, conversion of reservoir parameters to elastic parameters using a petroelastic model, seismic modeling, and interpretation of the results. Methodology Preparation of Reservoir Grids and Fluid Substitutions Reservoir simulation grids contain dead cells where there is no flow and no information on petrophysical parameters (saturations, porosity, for example). Before seismic modeling, dead cells must be filled with values for petrophysical parameters, in order to have realistic impedance contrasts during the seismic modeling step. Besides, dynamic parameters used in flow simulators (such as net to gross) have to be transformed in parameters used in petroelastical models such as Vclay. To populate reservoir grids, petrophysical parameters are upscaled: reservoir geologists often use cut-offs on logs and compute means of petrophysical properties from these logs for each facies. For seismic modeling purpose, realistic porosity and Vclay means must be recomputed for each facies group without cut-offs. Finally, a petroelastical model depending on porosity, Vclay, saturations and pressure is used to compute the elastic parameters (bulk density, P-wave velocity, S-wave velocity) in each cell of the reservoir model. This petroelastic model follows the Gassmann's theory, Castagna et al.5 for saturation effects. In absence of global theory to model the pressure effects, laboratory measurements are used.