Mineralogy based classification of carbonate rocks using elastic parameters: A case study from Buzios Field

Brazilian presalt carbonate reservoirs are highly heterogeneous. This feature is mostly justified by the nature of the original depositional system and subsequently diagenetic processes. Consequently, reserve estimates and production forecasting are under large uncertainties. In this geologic context, it is of great relevance to develop techniques that helps to obtain a detailed description on the spatial distribution of these different rocks. In doing so, it contributes to the understanding of presalt carbonate sedimentary deposits, providing inputs for more predictive reservoir models. Traditionally, these carbonates are grouped into three classes, from which only one exhibits reservoir properties. Using a dataset from Buzios Field, this work proposes a characterization of presalt carbonate reservoir rocks by grouping them in terms of their mineral composition. Taking advantage of rock physics concepts, we aim to potentialize the use of elastic parameters for multiple rock type discrimination. We explored several attempts for rock classification by using a Bayesian approach. Among all the tested propositions, a two-step workflow for five lithotypes classification, emerges as the most appropriate for the Buzios Field. In this scheme, three lithotypes represent good-quality reservoirs and the other two are low-porosity and Mg-clay-rich carbonates. The average root-mean-square error of the most likely a posteriori rock proportions is around 8.4%, only approximately 1% higher than the conventional three lithotypes configuration. To support that, we compared different methodologies for Bayesian classification at well-log scale through acoustic impedance and compressional to shear velocity ratio. Potential applicability of the proposed methodology at field scale is reinforced by similar results achieved using well-logs filtered to the seismic bandwidth. • Carbonate rock classification based on mineral proportions. • Influence of mineral composition on rock porosity and permeability. • Bayesian approach for multiple rock type prediction through elastic parameters. • Quantitative analysis of rock type prediction results at well-log scale.