Unlocking the properties of a presalt carbonate reservoir offshore Brazil with facies-constrained geostatistical inversion

Abstract

A continuing concern regarding presalt carbonate reservoirs offshore Brazil is how to derive accurate quantitative estimates of reservoir properties. It is challenging to understand the link between the facies model and the variation in elastic properties, recover a reliable model of elastic properties from seismic, estimate porosities and permeabilities to use in reservoir simulations, and ultimately close the loop in integrated geology and engineering workflows. This case study describes our use of geostatistical inversion as a tool to unlock reservoir properties. We show how the integration of diverse information from various sources and at different scales is used to produce a meaningful range of probabilistic realizations of this Brazilian deepwater presalt reservoir. We do this while respecting the reservoir properties observed at the locations of drilled wells. We also present a workflow for an optimized implementation of the inversion results at the modeling stage, resulting in fast and geologically consistent history matching in an extremely challenging reservoir management environment. In this way, we achieve accurate history-matched cases on a field level and on a well-by-well basis while remaining within the uncertainty limits. Therefore, we produce geologically plausible and reliable scenarios.

A facies‐constrained geostatistical seismic inversion method based on multi‐scale sparse representation

“…For the MCMC method, when there is no prior information constraint, the results may not match the actual geological conditions and sedimentary environments. Some researchers can reduce the multi‐solution of geostatistics by adding certain prior constraints to narrow the range of reservoir parameter solution space (Pacheco et al., 2022; Yu et al., 2020). Geostatistical seismic inversion uses the constraints of seismic data to control the solution space of reservoir parameters within the structural characteristics of seismic data, which can effectively reduce the multi‐solution of geostatistics (Azevedo & Soares, 2017; Grana et al., 2013).…”