A comprehensive framework for predicting elemental sulphur deposition in natural gas transmission systems via physics‐based simulation and machine learning surrogates

Abstract Elemental sulphur deposition poses a significant operational challenge in natural gas transmission systems, particularly downstream of pressure‐reducing devices where rapid pressure and temperature changes occur. These thermodynamic shifts reduce sulphur solubility, triggering nucleation and leading to issues such as valve malfunction, flow restriction, corrosion, and maintenance costs. This study presents an integrated physics‐informed framework to predict sulphur dropout within a pressure‐control valve. Unlike previous sulphur‐deposition models, the proposed framework integrates three‐dimensional computational fluid dynamics (CFD), equilibrium chemistry, sulphur solubility estimation, and classical nucleation theory within a spatially resolved workflow. CFD simulations resolve spatial pressure and temperature fields under varying operating conditions and valve openings. These fields are coupled with a chemistry module that adjusts equilibrium constants and solves a constrained reaction‐extent problem to estimate sulphur formation. Local sulphur solubility is determined through interpolation of literature data, while supersaturation‐driven nucleation rates are computed at the CFD cell level. The framework enables spatial prediction of sulphur supersaturation and precipitation within the valve. To support rapid screening, a random forest surrogate model was trained on simulation outputs, achieving a mean absolute percentage error of 15.4%, a root‐mean‐square percentage error of 20.4%, and an R 2 of 0.977. The framework provides a practical tool for identifying sulphur‐deposition hot spots, developing sulphur‐deposition operating envelopes, and enabling rapid risk screening through a machine‐learning surrogate.