ROCHA

The geologist, in the oil industry, has an important and challenging task: evaluating and correlating the large amounts of extracted data from oil and gas well exploration and production. One of the most important of these data sources is reservoir core. It is a sample of rock in the shape of a cylinder, taken from the side of a drilled well. This extraction is very expensive, and for that reason, it is done for just a few strategically selected wells.

The core analysis is used to obtain important information from the well, such as porosity, permeability, capillary pressure, fluid saturation, grain density, electrical characteristics, and others. All of these measurements help the geologists, engineers and drillers better understand the conditions of the well and its potential productivity. This analysis involves a step that classifies the rock by its lithology, in other words, into types and subtypes related to the geologic time.

Nowadays, the extracted core is examined manually by an expert geologist to make the lithological classification. This is a highly laborious and time-consuming process. Also, this analysis isn’t totally reliable, as it’s influenced by the geologist’s experience and by external factors such as illumination, rock torsions and fractures.

X-Ray computed tomography (CT) is becoming an excellent alternative tool to reservoir core analysis. Generally, medical CT scanners have been employed because of their availability and ease of use. Also, it’s a completely non-destructive means of examining the interiors of opaque solid objects. Therefore, the extracted images reveal a high-quality detail of the characteristics of a reservoir core that weren’t possible to see before. The CT scanner produces high-definition 2D images (called slices) that can be used to generate the mesh of the complete extracted rock.

ROCHA is a research and development project developed for Petrobras, which has the main objective of exploring and analysing reservoir core images produced by the medical CT scanner. Several different algorithms were used and evaluated to help the automatisation process in the lithofacies classification. As a result, a classification and a visualisation tool were developed.

The classification tool is based on Data Mining techniques, more specifically the clustering algorithms: K-Means, Adaptive K-Means, K-Means++, DBSCAN, Fuzzy C-Means, Competitive Agglomeration, Region Growing, MST Clustering and others. This tool is capable of determining the group’s tendency, the number of clusters in the data, and evaluating the results compared to the real lithologies.

The visualisation tool was created to be main information in the reservoir core’s classification. The software has different kinds of visualisations: volumetric, slices, longitudinal, and raw DICOM data.

Role Information:

In this project, my role was as Lead R&D Software Engineer. ROCHA was totally designed and developed from scratch from July/2011 to July/2012. As an R&D project, it offered significant challenges, such as creating a tool that could improve the quality and speed of a well’s productivity evaluation.

ROCHA presented excellent results at finding lithofacies with different clustering algorithms, selecting the one with better values of cohesion and separation of inner clusters.

The main technologies involved were: C++, Data Mining, QT, Image Processing techniques, OpenCV and Computer