AUTHOR=Liu Fan , Zhou Wensheng , Liu Bingxuan , Li Ke , Zhang Kai , Cao Chenming , Qin Guoyu , Cao Chen , Yang Renfeng 

TITLE=Flow Field Description and Simplification Based on Principal Component Analysis Downscaling and Clustering Algorithms

JOURNAL=Frontiers in Earth Science

VOLUME=Volume 9 - 2021

YEAR=2022

URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2021.804617

DOI=10.3389/feart.2021.804617

ISSN=2296-6463

ABSTRACT=In order to solve the complicated problem of flow field results in commercial numerical simulators, a method to describe and simplify flow field by using descending and clustering algorithms is proposed in this paper. At present, due to the serious heterogeneity of most water drive reservoirs, the streamline numerical simulation generated by the streamline distribution is complex. The streamlines in the three-dimensional streamline field interfere with each other and the connectivity between each well is not clear. Therefore, the advantages of streamline field are not obvious, which restricts the effective application of streamline numerical simulation. In order to improve the visualization degree of flow field and realize effective description of flow field, streamline field needs to be simplified. In this paper, on the basis of the extraction of streamline attributes, PCA dimension reduction algorithm is used to carry out characteristic dimension reduction for streamline attributes. Utilizing the principal component obtained after dimension reduction, the streamline field is analyzed by clustering. Based on the streamline characteristics, the streamline field is classified by clustering, and the mainstream line in the flow field is selected according to the clustering center obtained by clustering. The difficulty of describing and characterizing the flow field in detail by using streamline method is solved.