Reservoir microfacies analysis exploiting microscopic image processing and classification algorithms applied to carbonate and sandstone reservoirs

Two case studies are presented that apply a new quantitative micro-facies analysis method to multiple thin section photographs from the carbonate and sandstone reservoirs of two giant fields: (1) the Permo-Triassic carbonate reservoir formation from the South Pars gas-condensate field (offshore Iran); and (2) the Fasila sand reservoir of Middle Pliocene age from the Shah Deniz Field (offshore Azerbaijan). An essential part of comprehensive microfacies analysis of reservoir rocks is to reliably identify, characterize and quantify microscopic features including matrix components, mineral grains, rock fabrics, porosity, fossil contents and the products of diagenesis. This can be achieved by a combination of image processing techniques to assist in distinguishing paleoenvironments and depositional settings of porous and permeable formations. These techniques, applied to the case studies, are shown to effectively improve reservoir characterization for both carbonate and elastic reservoirs Image processing algorithms have to overcome several hurdles to achieve consistent and repeatable microfacies analysis and this is best conducted by exploiting several algorithms focus on specific recognition tasks. Initially, a histogram equalization algorithm is applied to adjust the quality of thin section images to make them consistently comparable with reference photographs. Microfacies analysis is then conducted on the adjusted images in steps to distinguish matrix texture, grain and pore-space boundaries and microfossil components. This is achieved by applying multiple functional image processing algorithms and sensitivity analysis sequentially targeting specific microfacies characterization objectives. Accurate grain size is quantified by a customized Graphical User Interface (GUI) algorithm. Pore detection and quantified two-dimensional porosity volumes are calculated for multiple thin-section images with a K-means clustering of two measurements extracted from color image space analysis. Color image data is also used to specifically distinguish the different minerals constituting the rock matrix, cement, and the form of the pore space in two dimensions. This enables these features to be consistently visualized and displayed in scatter plots that determine the precise microfacies to which each thin section belongs. Applying these techniques to the two case studies demonstrates how they can objectively and quantitatively provide microfacies analysis for carbonate and silicate reservoirs.