Reservoir Flow Field Characterization during Water Flooding: A Data Mining and Fuzzy Logic Approach

Area of Science:

• Petroleum Engineering
• Data Science
• Fuzzy Logic

Background:

• Accurate reservoir flow field characterization is vital for enhancing oil recovery in water flooding.
• Existing methods lack standardized criteria for selecting influential factors in flow field characterization.

Purpose of the Study:

• To introduce a novel quantitative framework integrating data mining and fuzzy logic for reservoir flow field characterization.
• To address the lack of standardized criteria in current methodologies for selecting influential factors.

Main Methods:

• Principal Component Analysis (PCA) and Spearman's rank correlation coefficient to identify key indices.
• A dual subjective-objective weighting strategy (Analytic Hierarchy Process and entropy weighting) for comprehensive index weights.
• Fuzzy logic and K-means clustering for quantitative characterization and flow field categorization.

Main Results:

• PCA and Spearman correlation identified oil surface flux, pressure, oil saturation, and permeability as critical factors.
• Dynamic factors were weighted higher (0.833) than static factors (0.167).
• K-means clustering delineated three flow field types with a silhouette coefficient of 0.724.

Conclusions:

• The data-driven framework provides a standardized, quantitative approach to reservoir flow field characterization.
• The methodology offers actionable insights for optimizing production strategies in mature, high water-cut oil reservoirs.
• This approach improves the efficiency of oil recovery in water flooding operations.