Combining Physics-based Model and Data Science for Ultra-deep Drilling
- Associate Professor Ryota Wada
Drill pipe dynamics of ultra-deep drilling
Ultra-deep drilling is expected to enhance both resource development and scientific research. When drilling with a long drill pipe, its dynamics cannot be neglected and the behavior of the drilling system gets complicated. A good understanding of the behavior is crucial for safe and efficient operation, but the monitoring data available during operation is limited. Our challenge is to tackle this problem by combining physics-based model and data driven model.
Our research aims to compensate for the shortcomings of physic al models by utilizing monitored data through deep learning. We incorporate engineer’s knowledge and physical understanding to build the data driven model, applied to gray box models and early detection of anomalies.