Estimation of drill cuttings particle sizes provides near real-time information that can be utilized in drilling operations. The analysis and interpretations of these rock samples is largely manual. Results are qualitative and can be inconsistent. We propose a practical automated method for estimating drill cuttings particle sizes using digital images. The proposed method is robust and combines several well-known and time-tested techniques like;; K-means clustering, numerically stable ellipse fitting, and standard morphological operations. Images are first preprocessed and clustered based on color values. The resultant grain mask is enhanced using binary morphological operations. Measurements are then extracted from the mask based on the major and minor axis fitted to the particles. The developed method is tested on drill cuttings digital images from four wells with varying lithologies and particle sizes. Sensitivity analysis on imperfect images shows the robustness of the methodology. The automated results validated against manual measurements show a difference of less than ten percent. Performance of the algorithm is in the order of seconds, which allows for its use in near real-time applications.
NSTL
