--> Metaheuristic Rock Property Determination Driven by Rock Type Constrained Global N-Dimensional Analysis

AAPG ACE 2018

Datapages, Inc.Print this page

Metaheuristic Rock Property Determination Driven by Rock Type Constrained Global N-Dimensional Analysis

Abstract

The simulation-to-seismic process leverages a petro-elastic model (PEM) to maintain the continuity of the simulator response with the seismic inversion. The requisite dry rock properties in the PEM are typically derived from measurements made on core or reference descriptions. A rock-type driven global dimensional optimization technique is employed directly on three-dimensional (3D) geocellular arrays to determine dynamic elastic properties from the reservoir simulator.

Rock typing by electrofacies leverages available well logs and reservoir description so that increased accuracy in describing rock-fluid interaction as well as the PEM itself can be achieved. To accommodate closed-loop rock property modeling in scenarios requiring dry rock property data, this paper introduces a global metaheuristic N-dimensional optimization method to compute elastic properties for each rock type based on constraints applied to seismic inversion.

Integrated rock typing and global N-dimensional optimization is used in a field case involving ill-posed petro-elastic characterization to develop a PEM for the simulator. This paper shows how global dimensional optimization can be accelerated by the inclusion of electrofacies defined as rock types, and then used to fine-tune the PEM for seismic driven model matching. Sensitivity parameterization is also considered to evaluate design variables of limited constraint and calibrate estimated rock properties through a blind test. The results demonstrate appreciable long wavelength similaritiesand relative solution improvement in the development of the PEM. . While local disparities exist, they are judged relative to the time of iteratively evaluating the pseudo-objective function and are deemed acceptable considering the tradeoff between workflow efficiency and solution accuracy.

This is the first known example of rock typing being used to accelerate a metaheuristic optimization process where the design variables and optimal solution are derived from three-dimensional arrays. While this process is examined specifically for PEM, it can be applied to other data limited modeling techniques.