Improved Formation Evaluation Through Principal Component Analtsis

Summary We investigate the potential usefulness of Principal Component Analysis (PCA) method in providing meaningful petrophysical information in the case of hydrocarbon exploration wells, in addition to the results obtained via conventional log interpretation, or to constrain and validate such results. PCA is a multivariate data dimensionality reduction technique, which may also be employed to discover and interpret dependencies and relationships possibly existing among the variables or to highlight dominant data trends. We applied PCA to geophysical logging data sets recorded in wells drilled in various oil and gas fields from Romania and evaluated the PCA results by comparison with production tests, core analyses, lithology logs and actual formation tops. This study suggests that PCA may successfully complement conventional formation evaluation methods. Straightforward applications of PCA include recognition and separation of lithostratigraphic units, reducing the uncertainly related to formation tops, and accurate delineation of reservoir intervals. Generally, the first principal components respond to major lithology changes, shale/clay content variations or significant borehole diameter variations. Higher-order principal components frequently reflect fluid-related data variability, but their use as hydrocarbon indicators or predictors for a certain area or structure requires a careful calibration by cross-checking with well test results and core analyses.