--> Multivariate statistical methods for characterizing Palaeozoic shale formations based on well logs and lab data

European Regional Conference and Exhibition

Datapages, Inc.Print this page

Multivariate statistical methods for characterizing Palaeozoic shale formations based on well logs and lab data

Abstract

The goal of the study was to enhance and improve information on the Ordovician and Silurian gas-saturated shales formations. Statistical methods were used for improving standard well logging interpretation. A comparative study was made between several different methods in order to check the validity of the proposed techniques. In this work authors focused on identifying the shale formations, especially the gas-bearing horizons, using multivariate statistical analysis, e.g. Principal Component Analysis (PCA), Cluster Analysis (CA) and Discriminant Analysis (DA). Additionally, the same set of logs was used in IPSOM (Techlog Software) analysis. This is an intelligent classifier to facies modelling that provides classification solutions based on clustering and the Kohonen algorithm (neural network technology). Along the classification, IPSOM provides probability of occurrence of predicted facies at each depth (presented as cumulative probability log) and combined curve of the highest probability of each group (probability of the winning facies). These curves along with statistics for each group are good QC tools for results evaluation. CA and DA were performed on well logs together with principal components (PC's).

CA was used for the classification and grouping data according to natural petrophysical features of rocks. Authors have applied several different clustering methods (hierarchical and non-hierarchical) and results based on 5 clusters were selected. IPSOM analysis confirmed the selection.

Groups corresponding to the gas-bearing zones were found (especially group 4 and 1 in Fig.1 (Go to Search and Discovery to view Fig. 1), depth intervals marked in yellow and green). Complex analysis showed diversification between shale gas formations and adjacent beds. It is shown that internal diversification in each gas formations is present.

It is concluded that multivariate statistical analysis appeared to be useful and quick tool for preliminary classification of members and gas-saturated zones identification. This approach offers possibility to define zones of interests based on the criteria parameters defined from well logs interpretation.

This study was financed in project MWSSSG Polskie Technologie dla Gazu Lupkowego.