Geoscience

Geologic property clustering

K-means analysis was used in combination with gridded geologic data to spatially classify areas of like rock. Final output is a grid layer that consists of cells that have all been classified according to their geologic properties. This approach can reveal geologic provenances that can be used to gain insight into engineering parameters regarding drilling, completion, and type curve development.

Fuzzy K-means applied to fracture data

Faults and fractures identified on a collection of resistivity image logs were used as input into a custom spherical K-means script. The resulting output included a collection of stereonet plots with individual fractures classified to specific groups. Membership values were included for each fracture to show strength of membership towards each group. This technique proved useful for identification of dominant fracture trends in various lithologic units.

3D geologic modeling (Petrel)

Geologic projects for the Utica and Marcellus have been built from the ground up and maintained as new data was acquired. Database management that included import and QC of approximately 1300 horizontal wells and 450 vertical wells. Petrophysical models including mineralogy, TOC, porosity, saturations, gas in place, and rock mechanics were utilized for building the 3D geologic models for both intervals of interest. Geologic models were structurally tied to all available vertical, horizontal, and seismic data. The models were subsequently utilized externally for completion and reservoir engineering modeling.