| Enhanced Oil Recovery (EOR) techniques are vitally important in the oil industry because these techniques could not only extend the life of wells, but also produce 10% to 30% additional oil from the reservoir. However, selecting the most suitable EOR techniques for unknown reservoirs is not easy for decision making. Based on literature, EOR screening criteria could help to find the best candidates for unknown projects, which is classified into two categories: conventional EOR screening and advanced EOR screening. In this research, an artificial intelligent (AI) method, hierarchical clustering algorithm, is adapted to analyze both steam flooding projects and worldwide EOR projects for the purpose of new steam flooding screening criteria and the prediction of EOR methods for unknown reservoir conditions.;Data pre-processing process were firstly conducted to ensure the data quality, then the hierarchical clustering algorithm was applied to the worldwide steam flooding projects and the worldwide EOR projects; after that the principal component analysis (PCA) was used to identify the major attributes in all clusters, and to visualize the projects in different clusters in a scatter plot by retaining high variance; and then descriptive statistics of using box plot and scatter plot were utilized to establish the screening criteria for each cluster.;Three uniqueness were illustrated in this thesis. First, detailed screening criteria has been established based on the hierarchical clustering results. Second, categorical features (formation type) was considered as one of the impact factors for clustering, which none of the existing advanced screening criteria methods included. Third, dimensionality reduction techniques have been applied successfully which clusters are clearly laid out in a two dimensional scatter plot. |
