Hadi Hajibeygi - TU Delft

Abstract

‘Mass-conservative multiscale simulation of 3D naturally fractured heterogeneous porous media (F-AMS)’

Natural porous media are typically described over large length scales with high-resolution (fine-scale) heterogeneous coefficients. In addition, many of them entail highly conductive narrow (lower dimensional) channels at multiple length-scales. Important is that our knowledge about the real-field formations is limited, and that we often work on ''several possible realizations'' than "the true one". As such, the linear systems are all uncertain and, thus, we often are interested in " acceptable, but efficient" solutions rather than exact (to the machine-accuracy) expensive ones. In summary, the result of these challenging characteristics lead to large (non)linear systems with high contrasts, which are beyond the scope of classical computational methods. To resolve this complexity, we present a novel mass-conservative Algebraic Multiscale Solver for flow in Fractured heterogeneous porous media (F-AMS). The multiscale formulation is developed for the coupled system consisting of n-dimensional heterogeneous matrix and (n-1)-dimensional narrow highly-conductive fractures. This is achieved by introducing a prolongation operator which consists of local basis functions for both fractures and matrix, with arbitrary level of resolution and coupling in both domains. While F-AMS is developed to provide mass-conservative approximate but efficient (with no iterations) solutions to the real-field applications, it can be used as an efficient preconditioner in a tow-stage scalable linear solver procedure. Several challenging test cases are presented to first benchmark our mass-conservative F-AMS, as a preconditioner in terms of CPU time against commercial multigrid (AMG) solvers, and then to demonstrate its applicability as a true multiscale solver (i.e., no iterations) for 3D real-field applications. Some existing challenges in this research field along with our ongoing research is also presented and discussed at the end.

Short biography:

Hadi Hajibeygi is assistant professor at the faculty of civil engineering and geosciences at TU Delft, where he leads a research group in Advanced Reservoir Simulation (DARsim) topics. He joined TU Delft 3 years ago from Stanford University, where he was a post-doctoral scholar. Hadi holds a PhD degree from ETH Zurich in 2011, and received ETH Medal of 2012 from the Rector of ETH Zurich. His research focus is on the development of next-generation reservoir simulators on the basis of multi-scale multi-physics computational framework.