- Contact Us
- Apply Now
Rock properties prediction using 3D X-ray computed tomography images
On behalf of the Department of Mathematics and Statistics in the College of Arts and Sciences, you are cordially invited to a seminar by Dr. Mohamed Soufiane Jouini from The Petroleum Institute, Abu Dhabi, UAE.
In the oil industry, rock properties obtained from cores, during drilling, are crucial to evaluate hydrocarbon reserves in oilfields. Properties are measured experimentally in laboratories on 1.5 inch diameter cylindrical samples. Nevertheless, measurement techniques might be time consuming in some cases or very difficult to carry out in unconsolidated rocks for example. We propose two new approaches to overcome these limitations by estimating rock properties using mathematical modeling and 3D X-ray Computed Tomography (CT) scanner images.
The first method is a stochastic approach where we model the textures of X-ray CT scan images by implementing a parametric model including first and second order statistics. Then, we use a neural network to find a potential relationship between textural descriptors and rock properties. After the learning process, the calibrated neural network allows predicting rock properties. The main objective is to obtain reliable continuous rock properties estimation values along cores. Also, we use the descriptors to classify the main representative textural facies in core samples before cutting core in order to optimize core plug extraction. The second method is a deterministic approach where we combine image processing techniques and numerical simulations to predict rock properties. The first step consists on segmenting 3D high resolutions micro-tomography images (1 to 40 microns) into solid and porous phases. Then based on the extracted pore network we estimate several properties like porosity, permeability and elasticity. We implement, numerical simulations techniques such as Lattice Boltzmann approach to simulate fluid flow and Finite Element Method to solve the elasticity equation on the digital image model of rocks. Finally, we compare estimation results with experimental measurements for real data using both approaches.
For more information please contact [email protected]