Leveraging Probabilistic MVCA of Well Logs for Defining and Quantifying Sweet Spots in Heterogeneous Reservoirs
Sweet spots are primarily recognized and categorized on the basis of porosity (Por) and water saturation (Sw), along with other factors (TOC, bed thickness and continuity, permeability, etc.). When sweet spots are comprised of vertically heterogeneous strata, quantification of Por and Sw can be problematic using traditional well log analysis methods. A major issue is the variation of matrix composition. Variable matrix grain density (via variable mineralogy and kerogen content) affects Por calculations, and variable matrix mineralogy (particularly clay mineral type and abundance) affects Sw calculations. These problems can often be reduced by leveraging the results of a probabilistic multivariate clustering analysis (PMVCA) using standard triple-combo logs as clustering variables. Simple multivariate clustering analysis (MVCA) can recognize and define interbedded disparate rock types (Electroclasses) that have different rock properties. But, the use of a Bayesian-based PMVCA instead of a simple MVCA provides a significant advantage because each sample (each digitizing step) is assigned probabilistically to each of several Electroclasses. PMVCA permits the development of a variable grain density profile characteristic of the matrix heterogeneity. This variable grain density profile can then be used to calculate a Por profile using a “modified” DPHI (density-porosity) equation within which the grain density is not a constant. In an analogous manner, it is possible to derive a Sw profile using a “modified” Archie equation where “modified” means that the a, m, and n values, typically treated as constants, are instead variables computed probabilistically from the different constant values assigned to each Electroclass. Also, if desired, different Rw values can be assigned to different Electroclasses. Thus, both Por and Sw profiles can be computed with more credibility than when using constant grain density values and constant Archie constant values. The process can be designed to automatically identify sweet spot Intervals in multiple well profiles (i.e. wells included in a given PMVCA) that satisfy designated “cutoff” criteria such as minimum Por, maximum Sw, and minimum Interval thickness. Although the procedure works best if inputs are calibrated using core data, it also can be used when no core data is available. The procedure works for both conventional and unconventional plays.
AAPG Datapages/Search and Discovery Article #90350 © 2019 AAPG Annual Convention and Exhibition, San Antonio, Texas, May 19-22, 2019