Incorporation of a priori information in reservoir history matching by regularization
Academic Article
Overview
Overview
abstract
A novel history-matching algorithm is developed on the basis of regularization, which is capable of incorporating a priori information about unknown reservoir parameters like porosity or permeability. The a priori information is considered to consist of order-of-magnitude point estimates which are directly measured from the core sample analysis, or are extracted from other field tests. The proposed algorithm proceeds as optimization of an objective function which is formulated by combining three indices of different nature: (i) the ordinary least-squares term which measures the deviation of the model output from the pressure observation data, (ii) the stabilizing term which measures the non-smoothness of the parameter, and (iii) another penalty term which measures the deviation of the estimates from the a priori point data. Simple rules-of-thumb are presented for selecting the two relative weights among the above indices sequentially in a stepwise estimation process. The proposed approach is compared with statistically-based methods of incorporating prior information with respect to the mechanism of constraining the parameter space. The proposed method is evaluated through numerical simulations on history-matching of a two-dimensional ideal gas reservoir.
