Guided Deep Learning Manifold Linearization of Porous Media Flow Equations for Digital Twins Operations

Explore advanced techniques in Digital Twin operations for closed-loop reservoir management in this 56-minute talk. Delve into physics-aware machine learning (PA-ML) enhanced reservoir simulation, focusing on two innovative approaches. Learn about improvements to the Embed to Control and Observe method (E2CO) for reservoir modeling, combining data-driven model reduction strategies with deep neural networks. Discover extensions to the POD model reduction technique, including state-space augmentation for bi-linear systems and autoencoder-based Koopman operator linearization. Understand how these methods can significantly speed up reservoir simulations while maintaining high accuracy in pressure and saturation predictions. Gain insights from Dr. Eduardo Gildin, a distinguished Professor of Petroleum Engineering at Texas A&M University, on cutting-edge research in reservoir simulation, optimization, and drilling modeling.

DDPS | Guided Deep Learning Manifold Linearization of Porous Media Flow Equations