A permeability model was developed for the reservoir within the Reservoir Characterization Project (RCP) study area at Postle Field, so that flow simulation could be performed. The biggest challenge was the heterogeneity within the reservoir, especially the presence of high permeability zones. The effects of these high permeability zones within a CO2 injection framework need to be analyzed in detail because the CO2 and miscible oil bank will flow through the path of least resistance and may cause early breakthrough and poor sweep efficiency. Permeability modeling based on multiple permeability distributions to characterize the Morrow A sandstone produced a more reliable reservoir model to simulate CO2 flooding within the fluvial system. The integrated permeability model was tested against a binary (sandstone – shale) model. A base case history match of liquid production (oil + water) was performed. This match showed that the performance of the integrated permeability model was better than the binary model with respect to matching early liquid arrival in specific wells High permeability zones within some wells were the cause of early fluid arrival and the integrated permeability model more accurately predicted these zones. An accurate characterization of these high permeability zones leads to a more reliable reservoir model for CO2 flow prediction.
Detailed understanding of the heterogeneities and complexity of reservoir architecture and flow properties are of utmost importance in the development and exploitation of commercial hydrocarbon reservoir. Characterization and simulation studies are performed on a continuous basis during the life of a field from initial exploration through appraisal, development and eventual abandonment. A key component of these studies is the knowledge of the reservoir permeability across the field. However, permeability data is sparse therefore; estimation methods have included empirical and statistical approaches, as well as the emerging pattern recognition techniques. The accuracy of most methods is greatly enhanced when the reservoir is subdivided into units with common flow properties (Facci, 2005) A Multiclass classification of reservoir facies based on Support Vector Machines (SVM) is used to identify sandstone petrofacies from core and log data (López and Davis, 2009). Each facies represents a specific flow unit. An SVM multivariate regression was used to predict permeability with the objective of making a reliable reservoir model to simulate CO2 flooding. The inclusion of high permeability zones within the reservoir model is of great importance and the integration of geophysical, geological and petroleum engineering data helps to constrain, and validate the results.
Postle Field, located in Texas County, Oklahoma, produces from the A sandstone member of the Pennsylvanian age Morrow Formation. The reservoir underwent water flooding since 1974, near shut-in in 1998, and CO2 flooding which began in late 2007. Average net pay thickness is 30 ft and it is interpreted as an incised valley. The production unit is the Hovey Morrow Unit (HMU) (Figure 1). These data are used as a constraint and validation tool for the permeability modeling of the A sandstone in order to predict CO2 flow.
