Exploration efficiency - Bru21
Exploration efficiency
Prototyping future geoscience data organization and analytics tools for improved exploration workflows
Interpreters in exploration assets need a productivity increase in their handling and integrating steadily increasing amounts of data. These terabytes of heterogeneous Big data, including various geophysical data ranging from 2D and 3D surveys with numerous processing versions and large populations of wells, need to be analyzed faster, yet reliably, to provide valuable inputs to decision makers. BRU21 aims at developing novel automated tools to increase data analysis efficiency in the exploration workflows through modern computational methods, e.g. machine learning and artificial intelligence, combined with cross-disciplinary subsurface expertise...read more
Program area team
- Kenneth Duffaut, Assoc. Prof. Geophysics and Rock Physics
- Carl Fredrik Berg, Assoc. Prof. Reservoir Engineering
- Frank Ove Westad, Adj. Prof. Big Data Cybernetics
- Ivan Tyukin, Adj. Prof. Artificial Intelligence, Machine Learning
- Veronica A.T. Caceres, PhD candidate
- Luc Alberts, PostDoc
- Kurdistan Chawshin, PhD candidate
- Pranav Audhkhasi, PostDoc
Program area manager:
Program area assistant manager:
Current Projects
The impact of well data quality on machine learning performance
PhD Candidate Veronica Torres Caceres
Main Supervisor Kenneth Duffaut
Sponsor: Aker BP
The project focuses on two topics:
1) Prototyping the “future” well database that integrates ‘’all” measurements acquired in wells together with their corresponding metadata;
2) applying and training machine learning algorithms to automatically access data quality, depth shifting, rock typing, similarity recognition, as well as estimate petrophysical and geophysical parameters
Project result: Automatic depth matching of well log data
Structured well log data base and algorithm for fast and accurate depth-matching of well log data
Automated seismic reconstruction of missing sections
Post Doctoral fellow Luc Alberts
Main Supervisor Kenneth Duffaut
Sponsor: Neptune Energy Norge AS
The project deals with designing and implementing software algorithms that automatically detect and trace geological unconformities typically found within seismic data sets, and label these before reconstructing the surfaces that have undergone uplift and erosion
Project result: Automated detection of geological unconformities
Machine learning-enabled workflow for automatic detection and classification of geological unconformities
Automated lithology classification employing whole core CT scans
PhD candidate, kurdistan Chawshin
Main Supervisor, Kenneth Duffaut
Sponsor: Equinor
This project aims at developing automated routines and workflows for lithology classification and estimation of transport properties. The main objectives of this project can be summarized as below:
- Enhance current utilization of whole core CT images in rock characterization workflows
- Rock typing based on automated image analysis routines
- Explore the application of machine learning algorithms to classify lithology
- Explore the application of machine learning algorithms to estimate transport properties such as porosity, permeability and water saturation based on the underlying lithology classification
Project result: Workflows to classify lithology using 2D and 3D CT images
Convolutional Neural Networks-based workflows for high-resolution classification of lithology and porosity