Understanding the physical properties of rocks is vital for hydrocarbon exploration and energy development. Traditionally, scientists rely on laboratory experiments that are expensive, time-consuming, and often limited by the availability of rock samples. But now, researchers are turning to deep learning and digital rock technology as smarter alternatives.
In a recent study, scientists developed advanced convolutional neural network (CNN) models to estimate key rock parameters such as permeability, porosity, tortuosity, and grain size. These factors are crucial in determining how fluids move through rocks—knowledge that supports energy exploration and efficient resource management.
By using digital rock models created from high-resolution micro-CT scans, the team trained deep learning algorithms, including state-of-the-art models like DenseNet201, ResNet152, MobileNetV2, InceptionV3, and Xception. The results showed that different models excelled at predicting specific parameters: CNN B was the best for tortuosity, Xception for porosity, and DenseNet201 for grain size.
This breakthrough highlights how artificial intelligence (AI) can speed up rock analysis, reduce research costs, and minimize reliance on physical sampling. Beyond energy exploration, such methods could also support sustainable resource use and help industries transition to smarter, cleaner technologies.
The research is strongly connected to the United Nations Sustainable Development Goals (SDGs), particularly:
- SDG 7 (Affordable and Clean Energy): improving efficiency in energy exploration.
- SDG 9 (Industry, Innovation, and Infrastructure): advancing digital technology in geology.
- SDG 12 (Responsible Consumption and Production): reducing material waste by minimizing physical sampling.
- SDG 13 (Climate Action): supporting sustainable practices in energy and resource management.
By merging geoscience with AI, this study shows how innovation can pave the way toward a more sustainable and energy-efficient future.
Source: https://www.scopus.com/record/display.url?eid=2-s2.0-85219504956&origin=resultslist
01/Ilkom/2025
