--> --> Abstract: Facies Characterization Using Multi-Resolution Graphical Clustering Techniques on the Pony Discovery (GOM) and MTJDA Fields (Malay Basin), by Stephen Carney, Grigoriy Perov, Kenneth Kemp, Ning Shin Ni Chai, Rick Beaubouef, C. R. Handford, and Ken Grush; #90124 (2011)

Datapages, Inc.Print this page

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

Facies Characterization Using Multi-Resolution Graphical Clustering Techniques on the Pony Discovery (GOM) and MTJDA Fields (Malay Basin)

Stephen Carney1; Grigoriy Perov3; Kenneth Kemp4; Ning Shin Ni Chai2; Rick Beaubouef1; C. R. Handford1; Ken Grush3

(1) E&P Technology, Hess Oil and Gas, Houston, TX.

(2) Belud Team, Hess Oil and Gas, Kuala Lumpur, Malaysia.

(3) Pony Team, Hess Oil and Gas, Houston, TX.

(4) U.S. E&P Reservoir Engineering, Hess Oil and Gas, Houston, TX.

Predicting reservoir facies in the subsurface is very important but can be challenging. Core data is a key component of many integrated work flows and helps calibrate log responses in various lithofacies. Typically however, core data is limited and interpreters are forced into interpreting facies based on cuttings, log signature and qualitative "jump correlation" from cored intervals. A relatively simple, user friendly and elegant solution to this problem is the application of a multi-resolution graphical clustering (MRGC) approach to generate facies (electrofacies). An MRGC approach has been successfully applied on low resistivity low contrast reservoirs in the MTJDA Fields (Malaysia) and complex facies analysis in the greater Pony area (GOM). Success was achieved on these highly complex problems due to the application of an integrated, multidisciplinary work flow and new technology.

Since the late 1950's a number of techniques have been developed to try and help the interpreter predict facies in non cored wells. These includes the use of log cut-offs, multi variant analysis, fuzzy logic and neural networks. The MRGC technique uses a coarse to fine self organizing map approach and is a non-parametric method which partitions data sets on the basis of their "data structure". According to their neighboring relationships, data points are clustered into "attraction sets" which are progressively and hierarchically merged into electrofacies. Some of the advantages of MRGC over some of the other techniques are that it has no operator bias, is reproducible, handles both continuous (log) and discrete (core) data and generates fit for purpose multi- resolution models.

The electrofacies logs are also ordered graphically in a geologically realistic manner which allows the interpreter to develop stratigraphic layering schemes based on stacking patterns in conjunction with other geological data. Electrofacies can also be combined into depositional facies or environments of depositon for inclusion into a 3-D reservoir model. They can also be "associated" with core lithofacies, this powerful tool relates lithofacies proportions observed in core to each electrofacies.

The MRGC approach is being incorporated in Hess reservoir description and modeling best practices. There is also great potential to further develop these innovative techniques on exploration and development projects in clastics, carbonates, conventional and unconventional reservoirs.