Abstract

As the reservoirs are becoming complex and thin, far beyond the seismic band width, there is a need to accurately map them and incorporate them into static models. In our study, to build a Field Development Plan, in a Meso-Cenozoic rifted basin, the main challenge was to delineate the lateral and vertical continuity between and within thin sand reservoirs. This basin was formed by the formation and development of the Central African Shear zone (CASZ). The obstacle was that the object model approach, based on present day analogue could not explain the hydrodynamic disequilibrium, evident from formation test data (MDT), in the same zone within a single structure. To address this challenge, a robust rock physics model was built using different probability density functions (PDFs) for individual zones, using petrophysical cut-offs and based on Bayesian classification. As seismic data is band limited, most of the thin sand reservoirs could not be individually resolved through seismic inversion and were picked up as packages of thin sands, clubbed along with shales, in the acoustic impedance volume. These predictions were limited to the resolution of the seismic data. Hence stochastic or geostatistical inversion was carried out using the well data and the deterministic acoustic impedance volume as a constraint. The PDFs, from rock physics model, were applied to the stochastic inversion results (stochastic acoustic impedance volumes), to build sand probability models for each reservoir zone. The final facies model, created using stochastic inversion, could effectively explain the subcompartmentalization of the reservoir zones, consistent with the formation test data and give a better history match. The multiple equiprobable acoustic impedance outputs and sand probability volumes, from stochastic inversion, were used for the uncertainty analysis of the facies model. In absence of pre-stack seismic data and shear logs, this approach can be effectively used to characterize a similar setting elsewhere.