--> --> ABSTRACT: Comparison of Multivariate Statistical Algorithms for Wireline Log Facies Classification, by Tang, Hong, Christopher D. White, M. Royhan Gani, Janok Bhattacharya; #90026 (2004)
[First Hit]

Datapages, Inc.Print this page

Tang, Hong1, Christopher D. White2, M. Royhan Gani3, Janok Bhattacharya3 
(1) Department of Petroleum Engineering, Louisiana State University, Baton Rouge, LA 
(2) Louisiana State University, Baton Rouge, LA
(3) University of Texas at Dallas, Richardson, TX

ABSTRACT: Comparison of Multivariate Statistical Algorithms for Wireline Log Facies Classification

Sedimentary facies are important in reservoir characterization because flow properties are commonly assigned using facies-specific correlations. In uncored wells sedimentary facies cannot be observed directly, and facies are inferred from petrophysical data. Three multivariate statistical methods (Previous HitbetaNext Hit-Bayesian, multinomial logistic regression, and discriminant analysis) are examined in this paper. The techniques are illustrated using log and facies data from a western African sandstone reservoir and from a shallow outcrop analog in the Wall Creek Member of the Frontier Formation, Wyoming, USA. Our new method uses empirical Previous HitbetaNext Hit distributions to model the distribution of petrophysical properties conditional to facies, eliminating difficulties in bin selection. Petrophysical property distributions are assumed conditionally independent, simplifying the use of Bayes rule. Confidence, discrimination, and probability logs compare the prediction performance of the statistical methods as well as illustrating influences of log combinations and sample size. Two-way analysis of variance compares prediction accuracy of the models. For a given dataset, there are no significant differences (with 90 percent confidence) in predictions by the three methods. Additional logs improve prediction accuracy from 30 to above 80 percent. Final prediction accuracy is 82 to 90 percent for these three algorithms. Including 10 to 20 percent of the complete core and facies data in model construction provides accurate predictions; models were validated against the data not used in model construction. The fitted classification models can generate three-dimensional log-derived facies distributions for geologic modeling and reservoir simulation. The three methods are straightforward, efficient, and have quantifiable prediction errors. Key words: Previous HitbetaTop distributions, Bayesian, multinomial logistic regression, discriminant analysis, facies classification, analysis of variance.


AAPG Search and Discovery Article #90026©2004 AAPG Annual Meeting, Dallas, Texas, April 18-21, 2004.