Permeability prediction by hydraulic flow units - Theory and applications

An improved technique based on the concept of hydraulic flow units or hydraulic units (HU's) is presented to calculate permeability distribution in uncored wells. Graphical probability methods, nonlinear regression, and the Ward's analytical algorithm are presented to perform cluster analysis on core data and identify prevailing HU's in a formation. A Bayesian-based probabilistic approach is discussed next to estimate HU's and permeability distributions in logged wells. This is an inverse problem that requires constructing an a priori training database to capture implicit relationships between core-derived HU's and various well log measurements. Information from the database is drawn statistically to infer HU distributions at target logged wells. The method is applied to two heterogeneous reservoirs, a carbonate formation, and a laminated sandstone formation. Comparisons of permeabilities calculated by the HU approach and other conventional techniques are provided that demonstrate the usefulness of the-HU method.