Control relevant complexity reduction for modeling of flow and transport in heterogeneous porous media Grant uri icon

abstract

  • In many instances, one is interested in simulating complex systems and also controlling and/or optimizing them to meet certain requirements. An example is to control injection rates in a CO2 sequestration in order to delay the CO2 plume formation or control remediation process in a bio-geo environment. In the case of oil and gas reservoirs, the idea of combining simulation, data management and optimization in a more structured approach has been evolving into what is known as closed-loop reservoir management. Although these systems have their unique characteristics and complexities, they share a common issue: large-scale simulation models. Mathematical models derived from first principles, conservation laws, and profound knowledge of materials physics are often obtained as the result of a discretization of a set of partial differential equations (PDEs). Hence, highly accurate and detailed descriptions of the underlying models induce dynamical systems of large dimensions either in the state or parameter spaces, as several millions of grid blocks are often necessary in the discretization process. The direct use of these highly resolved models for simulation is not generally feasible because their fine level of detail places prohibitive demands on computational resources, especially for real-time implementations. In this proposal, we will develop control relevant complexity reduction techniques that will allow reducing the complexity of the underlying system while honoring the input-output relations that are needed for the robust control.

date/time interval

  • 2015 - 2018