Developing a Digital Twin for Offshore Wells using Physics-Rooted Models

Abstract Accurate understanding of the physics of the wellbore and knowledge of production rates are essential as they serve as a key input to modelling and therefore affect results and the decisions made based on those. This paper presents a methodology to create a physical representation of the wellbore and to compute production rates from monitoring parameters utilizing physics-rooted models. When connected to real-time measurements, the process enables continuous production surveillance and integration into a digital oilfield solution. The approach is validated against data from literature and a real North Sea offshore field. This work consists of an integrated methodology using a mechanistic approach to replicate the physics of the wellbore. The process utilizes transient heat transfer calculation between a deviated wellbore and formation. Black oil models are used to determine the properties of the produced fluids, which may comprise mixtures of gas/oil/water. Basic fluid properties and static information including wellbore design are required for the initial model setup. The dynamic input comprises choke downstream pressure, choke valve setting, pressure and temperature at wellhead and downhole. Dynamic data may come from either SCADA (supervisory control and data acquisition) for near real-time calculation, or manual readings. The methodology is validated with two quality data points from various fields used by other authors such as (Hasan & Kabir, Fluid Flow and Heat Transfer in Wellbores, 2002), including an onshore and an offshore well. Moreover, the process is also tested against the publicly available historical dataset of the Norwegian offshore field Volve, which was in production from 2008 to 2016. This allows simulation of daily production rates throughout entire well life cycles. The simulation of the real field cases achieves an average error MAPE (mean absolute percentage error) of 11.75 % for the liquid rate. The novelty of this approach is the ability to run a digital twin of a wellbore based on data that is already acquired as part of standard well monitoring operations. Using the process as a VFM (virtual flow meter) can increase accuracy of production allocation and quality control (QC) physical MPFM (multiphase flowmeter). Having such a model-based approach offers significant potential for cost savings, for instance reducing OPEX (operating expenditure) by stretching physical metering cycles and lowering CAPEX (capital expenditures) by saving metering infrastructure.