A novel approach to designing wire mesh demisters for surface separators: experimental, CFD, and artificial intelligence perspectives

Multi-phase separators are crucial components in the petroleum industry as they separate different phases from produced fluids. One key element that influences their performance is the wire mesh demister. Typically, the design of these demisters is based on semi-empirical correlations, which may contain errors. This study involved designing and manufacturing wire mesh demisters using semi-empirical correlations. Subsequently, a gas-liquid flow loop was created in the laboratory, incorporating a horizontal gas-liquid separator. The wire mesh demisters were tested under various operational conditions within the laboratory separator, and the Stokes number in theory and experiment was measured to determine correction factors for use in the correlations. Due to limitations in water and gas flow rates in the laboratory, a CFD model was developed and verified using experimental results. After acquiring 76 simulation data points with the validated CFD model, a neural network was established to predict correction factors for the semi-empirical correlations. A new correlation was proposed for designing wire mesh demisters, with a mean square error value below 1.9%. The novelty of this study lies in introducing a new correlation for designing wire mesh demisters for field applications. [Received: November 19, 2023; Accepted March 3, 2024]