--> Abstract: Integrated Reservoir Modeling: Rapid, Repeatable Model Updates for 4-D Reservoir Surveillance, by Paula L. Wigley, Serge Galley, Andrew Evans, Roelof Pieterson, and Jim O'Connell; #90124 (2011)

Datapages, Inc.Print this page

Making the Next Giant Leap in Geosciences
April 10-13, 2011, Houston, Texas, USA

Integrated Reservoir Modeling: Rapid, Repeatable Model Updates for 4-D Reservoir Surveillance

Paula L. Wigley1; Serge Galley1; Andrew Evans1; Roelof Pieterson1; Jim O'Connell1

(1) Shell International Exploration & Production, Houston, TX.

As use of 4D seismic as a surveillance tool increases, integrated reservoir models which can be easily updated to reflect 4D observations become more important. In order to build a model which matches initial reservoir conditions and is tied to seismic it is essential to integrate all available geophysical, geological, petrophysical and engineering data.

To facilitate model updates, the reservoir models are built using workflows which remove interpreter subjectivity, ensure repeatability and reduce the cycle time between the seismic, the static model and the dynamic model. A Gulf of Mexico field is featured.

The starting point for each reservoir model is the pre-production seismic survey in depth domain with geobody extraction used for rapid delineation of flow unit distributions across the reservoir. By utilizing the seismic volume directly in the static modeling workflow a close tie between the model and seismic is maintained.

Upscaled well-log data together with 2D simulations determine the most appropriate layering scheme and result in a model built at the level of complexity required to address 4D results.

Rock types to be modeled are defined using measured permeability data linked with facies identified from core. The cored wells and associated rock typing are used to develop a neural network which automates facies identification in non-cored wells. This crucial step removes interpreter subjectivity and ensures repeatability of the model build.

Seismic attributes are cross-plotted against well data to ascertain the property, or properties, which can best be constrained by the seismic. This step, allows an improved definition of properties throughout the entire model volume rather than relying on well data alone. Additionally, the full 3D seismic volume is used rather than relying on average maps to guide property distribution.

Building the static model at simulation scale reduces the overall timeline, ensures compatibility of results and facilitates both static and dynamic model updates. The close ties between the seismic, static model and dynamic model usually result in a good history match without recourse to volume multipliers.

This integrated approach to reservoir modeling, together with the use of workflows maximizes integration of static, dynamic and seismic data; enables rapid model updates; removes interpreter subjectivity; ensures repeatability and reduces the iteration times from seismic to static model and subsequent simulation.