ErSE 300 Deep Learning in Subsurface Reservoirs

This course covers the fundamental concepts, models and algorithms of deep learning and their applications to subsurface reservoirs especially to hydrocarbon reservoirs. Reservoir characteristics and fluid flow are the main focuses of this course, which prepares students the required skills for compositional petroleum reservoir simulation and other related techniques including CCUS and ground water remediation. Current reservoir studies in all the Darcy scale, pore scale and molecular scale will be reviewed first to help students get trained in conventional reservoir studies. Mathematical foundations in popular deep learning methods will be introduced, including linear algebra, matrix decompositions, vector calculus, probability and distributions, and continuous optimizations. Representative deep neural network structure and algorithms for popular reservoir problems will be presented and explained to help students prepared for further research and work.




Basic numerical PDE courses, basic programming skills in MATLAB/Python, fluid mechanics, or consent of instructor.