Abstract
In this paper, a new workflow for data wrangling, cleansing, processing, automated identification and classification of drilling dysfunctions (e.g. bit balling, drillstring vibrations, etc.) for any number of wells is presented. Statistically-derived baseline rock strengths were used to improve drilling efficiency and transparency in completions design. The methodology leverages commonly available rig surface sensor data and mud motor parameters. Data handling and processing across multiple databases was automated. Time-series data was converted to depth-based data, and drilling dysfunction indicator metrics (mechanical specific energy - MSE, depth-of-cut, etc.) were computed. Subsequently, the combined data set was segmented based on formation tops, bottom hole assembly (BHA) used, and hole sections. A multivariate physics-based decision tree methodology was then applied for identification and classification of drilling dysfunction. A baseline MSE was statistically derived and a subsurface rock strength database was created for every formation in every well within a given prospect.
Over 100 historical unconventional wells across two shale basins were analyzed using the new workflow, and 10 different dysfunction categories were classified. The most commonly encountered drilling dysfunctions were whirl, stick-slip, and bit balling. An algorithm generates a set of reports per well based on independent segments for individual drilling operations, including strip charts of drilling parameters and dysfunction indicators, with identification of potential (low/high) problem areas. Auto-generated scatter plots, identifying optimal control parameters for every geological formation at a given location, are used for planning future wells and improving drilling performance. The calculated MSE baselines were averaged to build a repository of baseline MSEs per formation, which in turn can be utilized as a diagnostic tool for real-time MSE surveillance. The MSE baseline information can also be used as a representation of unconfined compressive strength (UCS), and thus used to formulate a reliable rock hardness map to enhance completion design.
Manual dysfunction analysis is time- and labor-intensive, and therefore often done either incompletely or bypassed altogether. Automating this process reduces time and effort significantly. The novelty of this work lies in the translation of the qualitative understanding of MSE to a reliable quantification of rock strength through the analysis of magnitudes and trends in drilling parameters and dysfunction indicators. Screening out drilling dysfunctions enables a much better rock strength approximation to aid completion optimization.
