Term 252

Automatic Classification of Drilling Particles Using AI and Data-Driven Techniques

Project Type: Self-Initiated

Project Description

‏This project develops an integrated on-rig system to automatically classify drilling cuttings and improve real-time formation evaluation. Traditional methods rely on manual and laboratory analysis, which are slow, costly, and inconsistent, limiting their usefulness during drilling operations. To overcome this, the system combines chemical conditioning to remove mud contamination with AI-based image analysis to accurately classify lithology types such as shale, sandstone, and carbonates. The results are then linked with drilling parameters to provide meaningful insights directly at the rig. A web-based interface delivers near real-time results, supporting faster decision-making by engineers and mud loggers. The system targets processing up to 1,000 images per hour, achieving at least 80% accuracy compared to well logs, and maintaining a low cost, offering a scalable and efficient solution for modern drilling operations.