oa Direct Conditioning of Gaussian Random Fields to Dynamic Production Data

In this paper, we address the conditioning of Gaussian random fields to time-dependent production data, e.g. pressure or production rate. The goal of obtaining a set of conditioned permeability fields is to make a prediction of the reservoir performance which includes a quantified uncertainty. Instead of the brute force approach of generating many permeability fields and keeping those that satisfy the production data within a give measure, we shall use a method that gradually modifies an initially generated permeability field towards a match with the production data, without destroying its local spatial structure.