A Novel Semi-Supervised Clustering Method for Carbonate Pore Size Distribution in Middle East Using Earth Mover Distance

Abstract

Pore space texture and its pore size distribution are critical parameters for displacement efficiency for waterflood, which have been applied in Middle East. For sandstone, pore size is a log-normal distribution usually. However, when it comes to carbonate reservoir, complex sedimentary and diagenesis yield complex pore size distributions including unimodal, bimodal and multimodal distribution. Therefore, it is not reasonable to classify different pore size distribution through statistic parameters such as mean, standard deviation, skewness and kurtosis. It is essential to propose a novel method to classify numerous pore size distribution results accurately and automatically by curve shape. This paper proposes the application of earth mover's distance (EMD) to pore size distribution, which is not mentioned before. The automatic clustering method is efficient and accurate to process massive data. Therefore, this method can prompt to systematic analysis and research of a giant scenario of reservoir injection characteristics in Middle East.