Estudo da razão de refino de malhas numéricas usadas na simulação de separadores supersônicos

Abstract

Supersonic separators constitute a promising alternative to conventional phase separation methods as they dispense with moving parts and external heating, reducing maintenance costs and energy consumption. However, experimental tests are costly, which makes computational simulations essential tools for optimizing these devices. The work fills a gap in the literature by comparing approaches for mesh construction for supersonic separators, investigating the influence of mesh refinement strategy on the accuracy and efficiency of supersonic flow simulations in a convergent-divergent nozzle. Two approaches are compared: uniform refinement in the axial and radial directions (Group 1) and exclusively axial refinement (Group 2). Five response variables are analyzed: shock wave position, minimum pressure, maximum velocity, minimum temperature, and mass flow rate. The results indicate that Group 1 presents greater accuracy, with lower values of the Grid Convergence Index (GCI) for all variables. Group 2, in turn, converges more quickly, with a total simulation time 24% lower, especially for less refined meshes. The mass flow rate presents the smallest errors (GCI < 0.1% in Group 1) as it is little influenced by compressibility; the minimum pressure proves to be the most sensitive variable, with errors up to 33% in Group 2 for coarse meshes. It is observed that uniform refinement is recommended when accuracy is a priority, while anisotropic refinement can be adopted in exploratory studies or under time constraints, provided that larger errors are tolerated.