Advanced Stress Estimation in Rock Formations Using Ultrasound and Machine Learning

Researchers at the University of Pittsburgh have developed a novel method for characterizing the stress state in subterranean rock formations. This method combines laboratory ultrasound testing with machine learning to accurately estimate stress based on ultrasonic wave velocities. By leveraging this innovative approach, the technology offers a more reliable and efficient way to understand stress distributions in rock formations, which is crucial for various applications in geomechanics and resource extraction.

Description

The invention involves sampling rock formations and testing compressional and shear wave velocities under varying triaxial stress conditions in the laboratory. The collected data is used to train a machine learning algorithm that can estimate stress states from wave velocity measurements. This method bypasses the limitations of traditional deterministic approaches, which often rely on elasticity theory and can lead to unreliable stress estimations. Instead, the machine learning algorithm directly relates wave velocity to stress state, providing more accurate results.

Applications

• Geomechanical analysis
• Resource extraction and mining
• Oil and gas exploration
• Earthquake and seismic research

Advantages

This method offers a significant improvement over traditional stress estimation techniques by using true-triaxial stress states and machine learning. It eliminates the need for intermediate steps that can introduce errors, providing a direct and reliable relationship between wave velocity and stress state. The technology is scalable and can be applied to various rock formations, making it a valuable tool for industries that require precise geomechanical analysis.

Invention Readiness

The technology has been developed and tested with in vivo data, demonstrating its feasibility and accuracy. Initial proof-of-concept studies have shown promising results, and the method has been validated through laboratory experiments. The research team is currently working on further development and refinement of the machine learning algorithm to enhance its predictive capabilities. Additionally, efforts are being made to expand the dataset by testing a wider variety of rock samples under different stress conditions.

IP Status

Invention