Integrating dynamic data into reservoir models - A multiple point perspective

The conventional approach to integrate production information in reservoir models is by iterative perturbation of the reservoir model until the production history of the reservoir is matched. A key drawback of these approaches is that the methods are amenable to random fields that are completely characterized by a two-point covariance function. In contrast, this paper presents two forward modeling approaches that investigate history matching within a multiple point modeling framework. In general, flow simulators transfer reservoir properties, such as permeability to well responses, such as pressure. In the first approach, this complex transfer function is approximated by a simpler proxy that represents the complex connectivity of the permeability field in the form of multiple point functions. This proxy representation is then used within an iterative Markov Chain method to obtain reservoir models that are conditioned not only to the available dynamic data but also preserve the correct connectivity characteristics of the reservoir. The complex patterns exhibited by geology can be represented in the form of multiple point statistics such as a multiple point histogram. In this second approach, the multiple point histogram corresponding to the prior model of the heterogeneity is perturbed until well responses are matched. This perturbation is accomplished by using a single parameter that is selected based on the deviation of the simulated from the observed well responses. Some preliminary results from this probability perturbation approach are presented for reservoirs exhibiting complex connectivity characteristics.