Python-based modelling improves torque and drag modelling in horizontal extended reach well

Abstract

Extended-reach drilling (ERD) wells are increasingly deployed to maximize reservoir exposure while reducing surface footprint, particularly in offshore environments. In Field X, typical ERD wells reach ~7,500 m measured depth (MD) with more than 3,000 m of horizontal departure, drilled from an offshore platform in 330 m water depth. These wells include two aggressive build-up sections - commonly 10-12 deg/30 m DLS - before entering long horizontal intervals drilled through alternating sandstone and shale. This complex geometry significantly elevates friction and drag, making accurate torque and drag (T&D) modelling crucial for successful casing-running operations, stuck-pipe avoidance, and overall well delivery reliability. To improve predictive accuracy for ERD completions, a Python-based workflow was developed to automatically extract, process, and analyze continuous hook-load data recorded by the mud-logging system. The workflow processed over 33,000 hook-load datapoints, applying rolling filters and peak-detection algorithms to generate high-resolution pick-up (PU) and slack-off (SO) profiles across the entire open hole. Validation against rig-measured loads demonstrated excellent agreement, achieving MAPE of 3.6% (SO) and 3.59% (PU), RMSE of 4.2 t and 6.5 t, and R2 values above 0.99, confirming the reliability of the automated extraction technique. Comparison with the original WellPlan model revealed that the existing T&D simulations over-predicted drag in the horizontal and shale sections of the ERD well, particularly across intervals with tight spots and micro-tortuosity. By recalibrating the model using Python-derived depth-based friction factors - ranging between 0.10 and 0.40, depending on lithology - the updated T&D model achieved strong alignment with actual tripping behavior. This improved calibration enhances operational safety margins, reduces the risk of excessive hookload or casing resistance, and strengthens planning for cementing, centralization, and completion installation. Overall, the results demonstrate that Python-based modelling provides a robust, scalable, and high-accuracy method for friction-factor calibration and torque-and-drag optimization in horizontal ERD wells, enabling more reliable and safer well-construction performance.