How Artificial Intelligence is Transforming Engineering Geology

Engineering geology focuses on studying and characterizing soils and rock masses as building materials for civil and mining engineering applications. While rock mechanics principles are well understood, the primary challenge of working with rock masses lies in their inherent variability and in the limited data available for accurate modeling (Starfield and Cundall, 1988). This challenge underscores the interdisciplinary nature of engineering geology, which has traditionally relied on manual processing and interpretation of diverse data sources to describe soils and rock masses.

Artificial intelligence offers a transformative approach to these challenges by leveraging large, integrated datasets to identify patterns and relationships that may not be apparent through conventional methods. By training algorithms on available geological data, machine learning models can enhance the potential of collected field data, help to identify geological features, and assist in decision-making processes with increased accuracy and efficiency. This capability is particularly valuable in rock mass characterization, geotechnical and geological modeling, and slope stability analyses.

The intersection of artificial intelligence and engineering geology represents a promising frontier. As we navigate increasingly complex geological environments and strive for greater safety and efficiency in operations, artificial intelligence is emerging as a crucial tool in our geological engineering toolkit, changing how we understand and work with rock masses.

1. The Growing Role of Artificial Intelligence in Engineering Geology

Engineering geology faces critical challenges that artificial intelligence is uniquely positioned to address:

• Data Integration: Modern engineering geology relies on vast datasets from drill cores, geophysical surveys, and remote sensing. AI can integrate these diverse data sources into coherent information, identifying patterns that might be overlooked through manual interpretation. This results in more accurate and data-driven decision-making.
• Automation in Geological and Geotechnical Analysis: Traditional logging and mapping methods are labor-intensive and prone to interpreter bias. AI applications, such as automated core logging and structural mapping, enhance the speed and accuracy of geological and geotechnical assessments. Machine learning techniques can recognize fracture networks, lithological variations, and geotechnical properties, reducing subjectivity and improving modeling.

2. ITASCA Innovations in Engineering Geology

Our consulting projects and practical applications have focused on key areas where artificial intelligence is transforming engineering geology.

2.1 Advanced Televiewer Analyses and Structural Logging

One of our primary focuses has been the development of sophisticated approaches to structural logging and evaluation. Our work (Owen et al., 2024) has demonstrated that machine learning and computer vision can significantly improve the accuracy and efficiency of structural logging. Through the integration of multiple data sources, including acoustic televiewer (ATV) and optical televiewer (OTV) data, we've developed systems that can:

• Automatically identify geological structures (Fig. 1).
• Automatically classify geological structures (open joints, close joints, faults).
• Reduce interpretation time while maintaining high accuracy.
• Provide consistent, unbiased feature classification.
• Generate structural statistics for integration into broader geological models.

The implications of these innovations for structural geology interpretation are profound, allowing engineering geologists to focus on interpretation rather than manual data processing.

2.2 Rock Strength Predictions

Our work integrating machine learning techniques with core logging data has improved rock mass understanding by filling in missing data on Point Load Index (Is50) (Thielsen et al., 2022).

By leveraging machine learning algorithms, we can better predict intact rock properties, obtaining new information for engineering geologists regarding:

• Fulfillment of missing data in core logging for geotechnical model construction.
• Key core logging quantities that show strong control on rock strength.

Together, these provide a better basis for developing more detailed geospatial models of intact and defected rock strength. ???

3. Future Developments and Emerging Trends: What We Envision

Looking ahead, ITASCA envisions several developments in the application of artificial intelligence to engineering geology:.

• Development of computer vision and machine learning models to assist engineering geologists in logging core samples.
• Development of automated systems for rock mass classification and analysis based on geophysical logs.

• Enhanced integration of multiple data sources (core logging data, geophysical logs, etc.) for more comprehensive structural understanding.

• Development of sophisticated surrogate models for complex geological systems.

4. Let’s Innovate Together

Integrating artificial intelligence into engineering geology represents a fundamental shift in how we understand and work with rock masses. Our consulting projects and practical applications demonstrate the enormous potential of these technologies to improve safety, efficiency, and accuracy.

As we continue to develop and refine these applications, we remain committed to pushing the boundaries of what's possible in engineering geology. The future holds exciting possibilities for even more sophisticated artificial intelligence applications, and we are proud to be at the forefront of this technological revolution.

For more information about our innovative approaches and how they can benefit your operations, please reach out to discuss how we can help address your specific geological challenges.

References

Owen, G., Furtney, J., Haywood, J., Thielsen, C., & Seifert, N. (2024). A machine learning approach to automating televiewer feature picking and classification. In Proceedings, Slope Stability 2024 (Nova Lima, Brazil, April 2024).

Starfield, A. and Cundall, P.. Towards a methodology for rock mechanics modelling. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, 25(3):99–106, 1988. ISSN 0148-9062. doi: https://doi.org/10.1016/0148-9062(88)92292-9. URL https://www.sciencedirect.com/science/article/pii/0148906288922929.

Thielsen, C., Furtney, J. K., Valencia, M. E., Pierce, M., Orrego, C., Stonestreet, P., & Tennant, D. (2022). Application of Machine Learning to the Estimation of Intact Rock Strength from Core Logging Data: A Case Study at the Newcrest Cadia East Mine. In Proceedings, 56th U.S. Rock Mechanics/Geomechanics Symposium (ARMA 2022, Santa Fe, New Mexico, June 2021), ARMA 22-283. Alexandria, Virginia: ARMA.