Computational Methodology to Study Heterogeneities in Petroleum Reservoirs

Characterization of hydrocarbon reservoirs is strategically important to define the productivity of oil and/or gas fields. It involves many challenges such as appropriate identification, classification and interpretation of diagenetic processes that directly affect the quality of the reservoirs. Proper studies demand integration and analysis of very large amounts of data, usually presenting high-dimensional feature spaces. Current methods have many manual steps leading to a limited exploration of the data. These challenges are being intensified due to the need of knowledge and time dedication from experts. We developed a novel methodology that combines established techniques, such as Principle Component Analysis (PCA), clustering methods, parallel coordinates and scatter plots, with features such as dynamic (magic) lenses – filter and shadow lenses –, axes reordering and color maps, to automatically perform reservoir characterization in order to assist the identification, validation and interpretation of petrofacies. Petrofacies is a set of petrographic characteristics of microscopic order which allow the analyst to understand the diagenetic processes, aiding in the evaluation of the potential for hydrocarbon storage in the reservoir. We have applied our methodology on several databases from different sedimentary basins – Espirito Santo and Parana basins (Brazil), Talara Basin (Peru) and Niger Delta Basin (Nigeria). We conclude that our method allows the analyst to gain insights about the entire database in a manner that is faster than the analysis using manual methods. It also allows validation of the results because it is a powerful tool that can qualitatively and quantitatively support the analyst in the identification, interpretation and validation of petrofacies. This new methodology can optimize data analysis of similar databases, accelerating the analysis and reducing the committed work by the experts.