DBPapers
DOI: 10.5593/SGEM2016/HB14/S01.003

A D-POSTERIOR STATISTICAL CLASSIFICATION PROCEDURE FOR THE PROBLEM OF RESERVOIR FACIES DETERMINATION

A. Zaikin, R. Salimov, I. Kareev, S. Usmanov
Friday 11 November 2016

References: 16th International Multidisciplinary Scientific GeoConference SGEM 2016, SGEM Vienna GREEN Extended Scientific Sessions, www.sgemviennagreen.org, SGEM2016 Conference Proceedings, ISBN 978-619-7105-80-3 / ISSN 1314-2704, 2 - 5 November, 2016, Book 1 Vol. 4, 19-26pp, DOI: 10.5593/SGEM2016/HB14/S01.003

ABSTRACT

The paper considers problem of estimation of porosity and permeability based on well logs. We suppose that linear regression for porosity and permeability holds true, but regression parameters are dependent on reservoir facies, whose location and number is unknown. We propose clusterization method of train data in order to determine reservoir facies location. Unfortunately, the same clusterization procedure can’t be applied to test data, which consists only of well logs. We therefore use Bayesian methodics for classification of test data as we suppose that facies properties are the same throughout all wells. One of these Bayesian methodics is a newly developed procedure of testing multiple hypotheses, which operates d-risks, special kind of Bayesian risks. Proposed classification procedure also uses some machine learning techniques. In result, our estimates have less deviation from porosity and permeability based on cores research than classical method of estimation.

Keywords: well log analysis, rock porosity, regression, Bayesian analysis.

PAPER DOI: 10.5593/SGEM2016/HB14/S01.003, A D-POSTERIOR STATISTICAL CLASSIFICATION PROCEDURE FOR THE PROBLEM OF RESERVOIR FACIES DETERMINATION

35 EURO ADD TO CART


Your cart ( items)

Subtotal:
Tax cost ():
Shipping cost:
Total:
Checkout Empty