Drilling Fluids Profiling—Is Your Mud Right for the Job?

Abstract

AbstractDrilling fluids are tested daily with many properties recorded to ensure that drilling operations are performed with the highest possible efficiency and in compliance with operator requirements. Mud check data are accumulated over the years but is heavily underutilised, as it is rarely used after the job is finished; this data are instrumental for optimization of future drilling fluids, however.The original global daily mud check data for ~30,000 wells (>780,000 individual checks) since 2016 is retrieved from the service company mud database. This large volume of data is prepared by selecting the key drilling parameters and fluid properties and performing the extensive cleaning to enable the usability of the data. It becomes evident that to make fluid profiling meaningful, just focusing on fluids properties is not enough, but the associated drilling parameters and conditions are critical for a meaningful analysis. The dataset is then evaluated using exploration data analyses and multidimensional statistical analysis techniques. An evolving barrel cost is calculated from product concentrations and product costs, which is then used to calculate the cost of dilution. This is combined with the costs of fluid treatment, circulation, and tripping to create a normalised cost of fluid use. By possessing a normalised cost of fluid use, a holistic cost model can be created an applied to future wells to better inform decisions on fluid selection.The clean dataset is comprised of more than 425,000 records with drilling parameters such as depths, hole size, bottomhole circulating temperature, and main drilling fluid properties, some being generic (e.g., rheology at different shear rates, fluid loss, and solids contents) and some specific for the fluid type, such as water-based mud (WBM) and nonaqueous fluid (NAF). As a part of the exploratory data analyses, all the attributes are evaluated by reviewing individual distributions, running pair-wise correlations, and performing cluster analysis to identify groups of highly correlated properties. The fluid types and families are then profiled using aggregated properties (5th, 50th, and 95th percentiles) and analysed for (dis)similarity between fluid types and families. Several dimensionality reduction techniques are also tested to visualise fluid similarities and cluster fluids by a multitude of properties.The developed analytics framework enables understanding of relationships between drilling fluid parameters and associated drilling properties, determines similarity of fluid systems in terms of key properties, and defines the fluid properties that are optimal or require adjustments for a set of drilling conditions. The paper explains how fluid profiles built on large-scale historical usage data enable new applications of fluid systems. To demonstrate the power of the fluid profiling approach, the analysis is performed for different geographical locations and drilling conditions. Examples also include product family similarity evaluation across fluid types, providing a path toward potential replacement of NAFs with similarity in properties but more cost effective and environmentally benign WBMs.The data analytics framework is developed and presented here, relying on hundreds of thousands of drilling fluid check records, cleaned and processed to enable data-driven decisions on drilling fluid type and family selection, and optimal fluid property ranges—improving drilling process efficiency.