Optimization and Reduced-Order Modeling of Geological Carbon Storage Operations

Lou Durlofsky
Stanford University

Geological carbon storage, in which CO2 is injected into depleted oil/gas reservoirs or brine aquifers instead of being vented to the atmosphere, represents a means of limiting greenhouse gas emissions. These processes can be optimized (based on economic and/or risk criteria) using computational procedures analogous to those applied for oil and gas production. In this talk, I will present methods and results for the optimization of carbon storage operations. The objective function we seek to minimize represents a measure of the risk of CO2 leakage through the cap rock at the top of the storage formation. Optimization and data assimilation are accomplished using (parallelized) derivative-free methods. Topics discussed will include the joint optimization of well placement and time-varying injection rates, optimization of brine-cycling events, optimization under geological uncertainty, leak detection based on inversion with above-zone pressure data, and the use of reduced-order (POD-TPWL) representations to model and optimize injection-stage operations.

Presentation (PDF File)

Back to Workshop III: Data Assimilation, Uncertainty Reduction, and Optimization for Subsurface Flow