Skip to main content Skip to secondary navigation
View of the front of Stanford Campus
Main content start
Closed-loop operations diagram for oil/gas and co2 storage.

Program Overview

Computational optimization, history matching (data assimilation), uncertainty quantification, and data interpretation are important technologies for modern reservoir management and carbon storage operations. The Stanford Smart Fields Consortium is a multidisciplinary program that performs state-of-the-art research in these important areas. We investigate a wide variety of approaches, algorithms and enabling technologies for optimization and history matching. A key SFC focus is the extension and application of computational methodologies developed for oil/gas production to carbon storage (and eventually hydrogen storage) operations. Several of our projects involve the development of deep-neural-network surrogate models to greatly reduce the computation required for Smart Fields applications. Many of our optimization and constraint-handling treatments are incorporated into a modular code base called the Unified Optimization Framework.

Download Program Overview

Smart Fields current and recent research areas include:

  • Wide variety of optimization techniques for well placement, well control, history matching, and
    closed-loop modeling of subsurface operations
  • Development of deep-learning-based surrogate models for flow simulation, with application to
    history matching and optimization
  • Deep-learning-based surrogate models for coupled flow-geomechanical simulation, with application to CO2 storage
  • Deep-reinforcement-learning for closed-loop reservoir management
  • Data science methods for interpreting permanent downhole gauges
  • Multifidelity approaches and error modeling for optimization and uncertainty quantification
  • Optimization of monitoring well location and type for CO2 storage
  • Treatments for handling constraints in CO2 storage optimization
  • Optimization under geological uncertainty
  • Data-space inversion for history matching and uncertainty quantification
  • Deep learning for geological parameterization for efficient/realistic history matching
  • Development of treatments for model error in history matching
  • Application of Smart Fields methodologies to practical CO2 storage operations

The Smart Fields Consortium will use and develop open-source software, and it is the intention of the Smart Fields Consortium that any software will be released under an open source model.

Recent Publications

Louis Durlofsky's Publications
Roland Horne's Publications

Please contact Samantha Mickens with any questions regarding SFC or the website.