Field Validation of a Full Transient Hydraulics Model Leveraging Digital Twin and Real-Time Rheology Technology, Case Studies from North Sea Continental Shelf

Abstract

Abstract Poor management of annular pressure regimes, or inadequate transport of drilled cuttings from bit to surface, results in significant non-productive time (NPT) and invisible lost time (ILT). On average one in three North Sea continental shelf wells requires the technical sidetrack of a section, pushing drilling projects over authorized expenditure (AFE). Insufficient control of real-time hydraulics is a significant contributing cause. Pre-well planning applications and real-time (RT) engineering applications have used, for decades, steady state hydraulics models. In recent times, this modeling has been augmented with transient hydraulic models which have improved the capability, especially in real-time. Our advisory and control system maintains a digital twin with tools and environment, updates the digital twin state from measurements and transient wellbore flow modelling, and assists range of operations. This paper details the features of a one-dimensional multi-phase transient model of wellbore flow capable of comprehensively resolving the temporal and spatial course of pressure, cuttings concentration in suspension, and those in static beds along the wellbore. The model considers local variations in rheology, cuttings slip, cuttings deposition, and cuttings mobilization. The cuttings concentration uses a drift-flux model, populated with coefficients from high-resolution 2D and 3D CFD simulations. The numerical solution is designed for effective real-time execution. A comparison of simulation outputs with measurements demonstrates the accuracy of the model: equivalent circulating and static densities (ECD & ESD) and standpipe pressures are compared to their measured counterparts from surface and downhole pressure sensors; simulated outlet cuttings mass is compared to measured skip weights. The evaluation also demonstrates that accuracy is a function of input data availability and quality: how RT rheological sensor data, providing a constant feed of Fann-75 data to the model, impacts simulation results. Results clearly show the superiority of a real-time transient hydraulic model, obtaining a 10-20% tighter correlation with measured ECD & ESD, and a simulated mass out of the wellbore within 4% of the measured skip weights during both drilling and section clean-up operations. Simulation results using input low-resolution bob-and-cup measured mud density and rheology have less accuracy than those supplied with a continuous feed of real-time density, rheology, and temperature from an automated measurement system. A sensitivity analysis on input variables serves to identify which fluid properties are the most essential to measure automatically. The paper will conclude with an overview on the implementation of the physics-based models in a SCADA style drilling advisory and automation system.