Abstract
A systematic method is proposed to better understand the relation between permeability, porosity, saturation, and clay content, to get representative permeability distributions as part of reservoir characterization.
This method utilizes the combination of a well-recognized permeability-porosity relationship and the Archie equation, as a predictive link between cores and well logs. This approach mathematically bonds the hydraulic (dynamic) and geometric (static) properties of the rocks. Permeability and rock types from well logs are obtained by careful theoretical and practical selection of parameters. The methodology converts a multidimensional correlation problem into sets of smaller correlations.
The procedure leads to a system of simultaneous equations, where the slope of one of the linear equations constitutes an exact calibration parameter between permeability and saturation. The equations allow a sequential transformation process from permeability-saturation, to permeability-porosity, and rock type-depth coordinates. The equations are solved for rock types, using reference grids constructed in all three coordinate systems. A calibration/diagnostic crossplot (slope-rock type) enables to find permeability, assess the integrity of the model, identify outliers, and perform consistency checks at different depths.
The practical implementation consists of transform tables, and correlations. Results can be extended to shaly sections, transition zones, and swept intervals. Some examples are presented using data from diverse reservoir classes.
Advantages:
Calculates permeability honoring and integrating conventional core analysis, capillary pressure, and logs
Requires log curves generally available in every well
Quantitative in uncored sections
Follows a deterministic method, so results can be easily extrapolated to uncored wells. No black box approach
Based on well-documented engineering and geological theory (flow zones, rock fabric, pore geometry).
Formulation permits a connection to the geology through rock fabrics, for interwell extrapolation.
Previous models lack a good nexus between cores and well logs. Neural networks, fuzzy logic, multiple regression, or a combination of statistical methods establish this connection, usually with difficulty and not very effectively. The creation of an effective direct link and accuracy are some of the strengths of the proposed procedure.
