Dimensionless data-driven model for optimizing hole cleaning efficiency in daily drilling operations

Poor cuttings transport in deviated wells limit drill rate, induce excessive torque and drag, or in severe cases result in a stuck pipe. This paper presents a generalized data-driven model that utilizes statistical techniques for optimizing hole cleaning efficiency under different drilling conditions in deviated and extended reach wells. For this purpose, the model is constructed based on three approaches including extensive experiments conducted in our flow loop of 5-m horizontal length (4.5in. x 2in.), a validated Computational Fluid Dynamics (CFD) model was developed, and experimental data were collected from the literature to develop a reliable predictive tool that can estimate cuttings concentration in deviated wells. The developed model utilized a non-linear regression method, and was trained with 75% of the gathered data and validated with the remaining 25% to ensure the capability of the proposed model for accurate estimation of cuttings accumulation under different conditions. Unique dimensionless parameters were developed to shift the model results from lab-scale to field-scale applications. Findings revealed that the developed model provides promising results in estimating cuttings accumulation in deviated wells (20-90 degrees from vertical). Predicted points lay in between 30% error margin in most cases, and the relation between estimated and measured cuttings accumulation has an adjusted R2 = 0.9. The proposed model outperforms the Duan, and Song models and introduces new dimensionless parameters to characterize hole cleaning efficiency during daily operations. The developed model proves to be a robust tool for simulating cuttings transport in real-time, monitoring cuttings accumulation, improving drilling efficiency, and avoiding Non-Productive Time (NPT) related to hole cleaning issues.