Shadegan oilfield is located in Khuzestan province in southwest of Dezful embayment, southwest of Iran. The latest 3D surface seismic was carried out in 1999 over an area of 192 km2 in this field to resolve structural ambiguities, delineate exact dimensions of the field, estimate reservoir parameters and perform a seismic facies analysis over the field for producer zones of Asmari reservoir.

Asmari reservoir in this filed consists of three producer zones including zones 2, 5 and 6. This paper will discuss about facies analysis of these zones based on 3D seismic multiattribute volume classification.

A sophisticated approach was undertaken to perform multiattribute seismic facies analysis in the Asmari reservoir for each zone. Many lithology sensitive attributes were extracted from the post-stacked 3D seismic data and some of them were rejected using a multidimensional correlation technique in order not to take part in classification algorithm and determining the optimum set of attributes to use in the classification. This helped with data redundancy problems inherent in multiattribute analysis and avoiding independent attributes by choosing an appropriate set of attributes for the classification. So, the seismic attributes used for the facies classification included Signal Envelope, Heterogeneity Seismic Amplitude, Integration Seismic Amplitude, Magnitude Seismic Amplitude and Volume Reflection Spectrum. The objective was to integrate the seismic attribute information through the use of an unsupervised clustering technique. Clustering techniques are widely recommended tools for classification issues.

The k-means algorithm is widely accepted as the standard technique of detecting different classes automatically from the measured data. This algorithm provided a facies database that was used to estimate lithofacies over the whole field. Using the clustering technique, three major lithologies including sandstone, carbonate and shale have been classified and displayed for each producer zone. Classified facies were calibrated to drilled wells and the results were confirmed.

Keywords:
machine learning, classification, centroid, Asmari reservoir, reference point, seismic facies analysis, Upstream Oil & Gas, continuity, facies, zone 2
Subjects:
Reservoir Characterization, Information Management and Systems, Geologic modeling, Seismic processing and interpretation, Artificial intelligence
Copyright 2007, Society of Petroleum Engineers
You can access this article if you purchase or spend a download.