Dépôt numérique

A texture synthesis based algorithm for simulating reservoir models.

Gardet, Caroline, Le Ravalec, Mickaele et Gloaguen, Erwan (2014). A texture synthesis based algorithm for simulating reservoir models. In: 76th European Association of Geoscientists and Engineers Conference and Exhibition 2014: Experience the Energy Incorporating - SPE EUROPEC 2014, 16-19 juin 2014, Amsterdam, Pays-Bas.

Ce document n'est pas hébergé sur EspaceINRS.


The geological complexity of reservoir calls for multiple points statistics to represent curvilinear objects, which cannot be done with usual two-point statistics. Computer graphics algorithms and geostatistical multiple point algorithms share many ideas. They both refer to a training image depicting the expected geological architectures and produce images with the multiple point statistics inferred from the structural content of the training image. The difference is related to the ultimate purpose. In geostatistics, we aim to build realizations conditioned by hard data. In computer graphics, the idea is to reproduce textures that cannot be distinguished from natural textures by human eyes. Here, we present an texture synthesis algorithm modified to properly handle geological reservoir simulation. Various tests are performed to investigate how this approach can help reproducing geological objects (with similar shapes, connectivity, density, distribution, etc.). A few improvements are suggested to make the simulation algorithm more efficient and appropriate for reservoir modeling. We also tried to decrease the CPU-time without degrading the simulated image. Last, we present a few numerical tests focusing on training images describing porous media as well as geological formations with channels or faults.

Type de document: Document issu d'une conférence ou d'un atelier
Mots-clés libres: computer graphics; geology; petroleum reservoir evaluation; petroleum reservoirs; porous materials; textures
Centre: Centre Eau Terre Environnement
Date de dépôt: 10 mai 2018 15:51
Dernière modification: 10 mai 2018 15:51
URI: https://espace.inrs.ca/id/eprint/3760

Actions (Identification requise)

Modifier la notice Modifier la notice