A method for automatic history matching of a compositional reservoir simulator with multipoint flux approximation |
| |
| Authors: | Dmitry Eydinov Sigurd Ivar Aanonsen Jarle Haukås Ivar Aavatsmark |
| |
| Institution: | (1) Centre for Integrated Petroleum Research (CIPR), University of Bergen, Allégt. 41, N-5007 Bergen, Norway |
| |
| Abstract: | A method for history matching of an in-house petroleum reservoir compositional simulator with multipoint flux approximation
is presented. This method is used for the estimation of unknown reservoir parameters, such as permeability and porosity, based
on production data and inverted seismic data. The limited-memory Broyden–Fletcher–Goldfarb–Shanno method is employed for minimization
of the objective function, which represents the difference between simulated and observed data. In this work, we present the
key features of the algorithm for calculations of the gradients of the objective function based on adjoint variables. The
test example shows that the method is applicable to cases with anisotropic permeability fields, multipoint flux approximation,
and arbitrary fluid compositions. |
| |
| Keywords: | History matching Parameter estimation Inverse problems Numerical optimization |
| 本文献已被 SpringerLink 等数据库收录！ |
|
