Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn how deep neural networks can revolutionize the modeling of complex fluid flow processes in subsurface rock formations through this 53-minute seminar presented by USC associate professor Birendra Jha. Explore the critical challenges in subsurface fluid dynamics relevant to enhanced oil recovery, groundwater remediation, waste disposal, and CO₂ sequestration, where traditional physics-based simulations are computationally expensive and time-consuming. Discover how viscosity and density contrasts between fluids create viscous fingering patterns that significantly impact flow behavior, mixing processes, and chemical reactions in underground reservoirs. Understand the geomechanical complications that arise when fluid injection induces stress, deformation, and potentially seismic activity, leading to storage leakage and reduced efficiency in carbon sequestration projects. Examine cutting-edge deep learning approaches including Fourier Neural Operator (FNO), DeepONet, Vision Transformer (ViT), and U-Net architectures that serve as surrogate models for predicting viscous fingering patterns, CO₂ leakage scenarios, and ground deformation. Gain insights into how these neural network models are trained on physics-based simulation data to rapidly and accurately extrapolate predictions to new initial conditions, boundary conditions, and rock-fluid property combinations where traditional simulators prove too slow or unreliable. Learn from an expert who combines academic research with extensive industry experience at Schlumberger, Occidental Petroleum, and iReservoir, and whose work spans geomechanics and multiphase flow applications in CO₂ sequestration, geothermal energy systems, and hydrogen generation and storage technologies.
