双水平集逼近油藏模型特征的数值模拟方法

Abstract

In this paper, Uzawas algorithms by steepest gradient descent scheme and operator splitting scheme for dynamic reservoir characterization by a binary level set approach are explored. The permeability of two-phase porous media flow can be recovered by the augmented Lagrangian optimization method with well data and seismic time-lapse data. After adding the constraint functions to the augmented Lagrangian optimization functional, the saddle points can be found by steepest gradient descent scheme and operator splitting scheme, which is based on the essence of binary level set method with few difficulties. In the end, the numerical examples demonstrate that the given algorithms will be efficient, stable when the permeability is recovered.