Multi-level parameter structure identification for two-phase porous-media flow problems using flexible representations

2009 ◽  
Vol 32 (12) ◽  
pp. 1777-1788 ◽  
Author(s):  
Inga Berre ◽  
Martha Lien ◽  
Trond Mannseth
2020 ◽  
Vol 144 (4) ◽  
pp. 449-492
Author(s):  
K. Mitra ◽  
T. Köppl ◽  
I. S. Pop ◽  
C. J. van Duijn ◽  
R. Helmig

2000 ◽  
Vol 16 (2) ◽  
pp. 315-332 ◽  
Author(s):  
Geir Nævdal ◽  
Trond Mannseth ◽  
Kari Brusdal ◽  
Jan-Erik Nordtvedt

Mathematics ◽  
2019 ◽  
Vol 7 (5) ◽  
pp. 412 ◽  
Author(s):  
Min Wang ◽  
Siu Wun Cheung ◽  
Eric T. Chung ◽  
Yalchin Efendiev ◽  
Wing Tat Leung ◽  
...  

In this paper, we propose a deep-learning-based approach to a class of multiscale problems. The generalized multiscale finite element method (GMsFEM) has been proven successful as a model reduction technique of flow problems in heterogeneous and high-contrast porous media. The key ingredients of GMsFEM include mutlsicale basis functions and coarse-scale parameters, which are obtained from solving local problems in each coarse neighborhood. Given a fixed medium, these quantities are precomputed by solving local problems in an offline stage, and result in a reduced-order model. However, these quantities have to be re-computed in case of varying media (various permeability fields). The objective of our work is to use deep learning techniques to mimic the nonlinear relation between the permeability field and the GMsFEM discretizations, and use neural networks to perform fast computation of GMsFEM ingredients repeatedly for a class of media. We provide numerical experiments to investigate the predictive power of neural networks and the usefulness of the resultant multiscale model in solving channelized porous media flow problems.


Author(s):  
Y C Lee ◽  
P H Gaskell

A fast and robust multi-grid algorithm for the efficient solution of diffusion-like, elliptic problems which exhibit strong discontinuous jumps in diffusivity is presented. Although generally applicable to this class of problem, the focus for illustrative purposes is that of porous media flow; in particular, such flows for which accurate solutions can only be achieved if the full permeability tensor is taken into consideration. The merits of adopting one or the other of two different approaches to deriving a discrete analogue to the steady-state Darcy equation, namely a novel weighted average of permeability formulation and a continuity of flux preservation method, are explored. In addition, automatic mesh refinement is incorporated seamlessly via a multi-level adaptive technique, making full use of the local truncation error estimates available from the inclusive full approximation storage scheme. Adaptive cell- and patch-wise mesh refinement strategies are developed and investigated for this purpose and used to solve a sequence of benchmark problems of increasing complexity. The results obtained reveal: (a) the ease with which the overall approach deals with generating accurate solutions for flows involving both distributed anisotropy and strong discontinuous jumps in permeability; (b) that both discrete analogues produce equivalent results in comparable execution times; and (c) the significant reductions in computing resource, memory, and CPU, to accrue from employing automatic adaptive mesh refinement.


Sign in / Sign up

Export Citation Format

Share Document