Biography
Marcus is a Computer Scientist who graduated at Universidade Federal de Lavras (Brazil). He holds a Ph.D in Computational Physics from ETH Zurich (Switzerland). Later, he worked in an industry-based research project at ORG-Geophysical (Norway) through a Marie-Curie (ITN) Fellowship, which focused on electromagnetic methods for hydrocarbon exploration. His general research interests are numerical methods, scientific and high performance computing applied to geosciences.
Within iCRAG he develops a computer software capable of generating artificial representations of the subsurface. These geo-model representations identify and reproduce the connectivity between the reservoirs. Accurate artificial reservoir models support geologists and engineers on the decision processes of drilling projects to maximize oil recovery. The main question that is being asked in that project is: Are inter-connections between multiple reservoirs important in oil recovery?
Project title: Hierarchical Compression modelling for Petrel.
Technical description
The goal of this project is the software development of a Plug-in in Schlumberger Petrel using Ocean API that generates geologically more realistic models of specific reservoirs. Our research identifies the significance of bed scale connectivity measures, a factor that is not generally recognized or incorporated in reservoir modelling approaches. Reservoir connectivity is created in erosion events, where a sandstone bed base erodes into another underlying sandstone bed. The Amalgamation Ratio characterizes the sand bed erosion on a vertical sample line. The algorithm uses object based techniques to model facies distribution with Net:Gross (NTG) and Amalgamation Ratios (AR) used as inputs. The compression method, which is the algorithm central to the sedimentological modelling implemented in the software, uses the input to compress or expand the objects to achieve the target parameters NTG and AR honouring reservoir connectivity and the geological hierarchy. A recursive approach is used to stochastically distribute self-similar objects within a higher-level object. Further to the distribution, gridding techniques must contextualize the small objects according to the geology of the big object container.
Read more here.
Role
- Postdoctoral Researcher
Institution
- UCD
Research Area
- Earth Resources
Expertise
- Reservoirs and Storage
