Integration of Cluster Analysis and Rock Physics for the Identification of Potential Hydrocarbon Reservoir

Rock physics has proven its credibility for the quantitative seismic interpretation of reservoir and reservoir characterization. In this study, we implemented K-means clustering to group data points more objectively and without prior misconceptions into clusters. Furthermore, we used the grouping suggested by the algorithm for our study. We integrated rock physics and K-means cluster analysis to determine a possible hydrocarbon reservoir using three well logs data. Initially, K-means clustering was implemented on density logs and based on the arithmetic mean of each density log cluster, sandstone and shale-dominant parts were identified and separated. Then, rock physics parameters were computed for the sandstone-dominant part of the well logs. Based on the cross-plots of Lame's constants with density product, and Lame's constants ratio, additional lithology discrimination was done for the identification of a clean gas-sand zone. In the clean gas-sand zone, two different pore fluids zones were identified based on V-p/V-s ratio, Poisson ratio, P- and S-wave impedances. Finally, statistical analysis was carried out to observe the underlying frequency distribution of rock physics parameters. The integration of cluster analysis and rock physics gave us significant information about the presence of different fluids and the existence of a potential hydrocarbon reservoir.