Condition object- or event-based reservoir models using multipoint statistical simulations

Abstract

A computer-based method of conditioning reservoir model data includes performing a modeling process within a 3D stratigraphic grid to generate an initial model including one or more facies objects within a model volume, the modeling process including parameter distributions , initial and boundary conditions, and deposition and erosion events to define facies objects within the model volume. A local variable constrained model is computed using the mismatch between the initial model and conditioning well data and the underlying input trend model. The method also includes performing a multi-point statistical simulation with the constrained model that is fully constrained by the initial model at locations where the initial model is consistent with known well data and potential input trend models and where the initial model is not The location of the matching known well data or potential input trend model is not varied between the constraints of the initial model to allow compliance with known data.

Description

technical field [0001] The present disclosure generally relates to computer-based modeling of physical properties. In particular, the present disclosure relates to tuning object or event based reservoir models using local multipoint statistical modeling. Background technique [0002] The goal of reservoir modeling is to build a 3D model of petrophysical properties (usually porosity and permeability, and sometimes water saturation) that reservoir engineers can use to run flow simulations, predict future oil and gas production, and ultimately recovery factor, and design well development plan. In most geological settings, especially clastic ones, porosity and permeability heterogeneity is primarily driven by facies depositional events. Therefore, the spatial distribution of porosity and permeability can be characterized primarily by the geometry and location of facies geologic bodies such as meandering sand channels. So geological modelers often first build 3D facies models ...

AI Technical Summary

Problems solved by technology

However, both of these methods have the disadvantage of being poorly correlated with conditioning data, since such conditioning data are drawn from real-world observations at specific well locations and may not correlate well with the user-specified geometry of the object to be simulated match the size

Even when the well data is perfectly consistent with the object geometry, object- and event-based approaches often stop abruptly and fail to fully match the well data due to the large combinatorial space of all possible object configurations within the reservoir model

Images