Integrated Reservoir–Wellbore–Choke Coupling Model for Deep Coalbed Methane

Deep coalbed methane (CBM) reservoirs exhibit ultralow permeability, high in situ stress, and pronounced stress sensitivity. The resulting feedback between reservoir deliverability, wellbore liquid transport, and surface choking cannot be represented reliably by isolated reservoir or wellbore calculations. This study develops an integrated reservoir–wellbore–choke coupling model for deep CBM wells. The reservoir submodel adopts a dual-porosity, single-permeability formulation for matrix-to-natural-fracture mass transfer, incorporates hydraulic fractures through non-neighboring connections, and accounts for Langmuir adsorption/desorption and effective-stress-dependent permeability. Gas–liquid flow in the tubing or annulus is calculated with the Beggs–Brill correlation, whereas critical and subcritical flow through the wellhead choke is evaluated with the Sachdeva mechanistic model. Bottom-hole flowing pressure (BHP) serves as the coupling variable in a partitioned sequential-iterative scheme. For each time step, the reservoir model predicts gas and water rates at a prescribed BHP; these rates are passed to the choke and wellbore models, whose returned BHP updates the reservoir boundary until convergence. Newton iterations solve the reservoir equations, and the critical liquid-carrying rate identifies the end of stable natural flow and the onset of liquid-loading risk. Application to Well H1 yielded agreement scores of 80.99%, 79.93%, 94.17%, and 90.48% for the gas rate, water rate, BHP, and wellhead tubing pressure, respectively, with an overall mean of 86.39%. Gas content governed the mid- to late-time deliverability, while tubing and choke sizes controlled the trade-off between friction loss, drawdown, and liquid unloading. A 2–3/8 in tubing string combined with a 12 mm choke provided the most balanced performance. The model supports life-cycle production forecasting and integrated completion and production optimization for deep CBM wells.