Authors:

Luisa Sanchez

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

Precise construction of stratigraphic columns is essential for geological interpretations, especially in identifying potential $CO_2$ reservoirs, which are crucial for mitigating climate change by sequestering carbon dioxide. A detailed record of rock layers is critical for understanding subsurface geology and making informed decisions about resource extraction and reservoir monitoring. The lack of detailed data on rock arrangements presents a substantial challenge for the geological characterization of potential $CO_2$ reservoirs. This project addresses these challenges by employing machine learning techniques to enhance lithofacies classification and prediction using well logs from the Illinois Basin. Lithofacies are rock units characterized by specific combinations of lithologic and sedimentologic properties that distinguish them from other units. To identify the lithofacies within the well, the model utilizes a Multilayer Perceptron to integrate data such as effective porosity, permeability, shear wave velocity, compressional wave velocity, density, and the Vp/Vs ratio. By detecting significant changes in the well log parameters, the model classifies these variations into specific lithofacies. The project aims to create a 2D map simulating the stratigraphic column of the area, providing valuable insights for $CO_2$ storage site identification. To ensure model accuracy, cross-sections from existing maps of Illinois will be used to validate the presence of at least four of the ten identifiable facies. The comparison of experimental and simulation results supports the integration of these techniques into geological workflows. Further research is recommended to explore additional geological attributes and machine learning algorithms to enhance model robustness and applicability in various geological settings, with a particular focus on identifying suitable $CO_2$ reservoirs. Keywords: Lithofacies Classification; Stratigraphic Column; Machine Learning

Source:

Purdue University / 2024

Topics:

No topics listed

Co-authors:

Luisa Sanchez