作者: Xiaoyi Liu , Quanlin Zhou , Jens Birkholzer , Walter A. Illman
DOI: 10.1002/WRCR.20489
关键词:
摘要: [1] Reduced-order models (ROMs) approximate the high-dimensional state of a dynamic system with low-dimensional approximation in subspace space. Properly constructed, they are used to significantly reduce computational cost associated simulation complex systems such as flow and transport subsurface. A key component model reduction is construct where we look for solutions. In this work, apply inverse modeling use solution parameter space underdetermined geostatistical problems which seek solutions any given parameters needed inversion process. The constructed by collecting variable (e.g., pressure) distributions domain. Each distributions, called snapshots, contains result full forward test basis vector input parameters. We then linear combinations snapshots modeling. modeling, spanned cross-covariance measurements parameters; hence, name ROM reduced-order (GROM). also show that minor loss accuracy model, estimation still high, saving significant, especially large-scale number unknowns enormous.