Abstract: The lithology of conglomerate is rich and variable, the pore structure is complex, the classification of reservoir type is difficult, without standardized evaluation parameters. Taking the conglomerate reservoirs of the Upper-Lower Karamay Formation in District Min-7 of Karamay Oilfield of Xinjiang as the object of study, based on the high-pressure mercury injection data of the reservoir rocks of the Formation 106, the classification scheme of the conglomerate reservoirs was established by using the clustering analysis method; the classification parameters of the conglomerate reservoirs were simplified and clarified by the methods such as the variance calculation and the validity of this method was verified by the use of the discriminant analysis in the end. The results show that the non-homogeneity of the conglomerate reservoir in the study area is strengthened as the physical properties become better and the pore-throat size becomes larger, reflecting the complexity of the conglomerate pore structure; the clustering analysis after data preprocessing can effectively classify the reservoirs, and there are significant differences between the classes; based on the quantification of parameter centralization and dispersion, six parameters for reservoir classification, such as the median pressure and the average pore-throat radius, are preferred; after the validation of the discriminant analysis, the accuracy of classification is still as high as 95.80% after the parameter preference, indicating that the method has high classification accuracy and is not geographically restricted, which makes it valuable for generalization.

Key words: Karamay Formation, conglomerate reservoir classification method, pore structure, clustering analysis, discriminant analysis, parameter preference