توصيفگر ها :
عدد رينولدز بحراني , ريزش گردابه , سيلندر مستطيلي , گوشه گردشده , نسبت منظري , عدد رينولدز , كنترل غيرفعال جريان
چكيده انگليسي :
The flow and heat transfer around two-dimensional rectangular cylinders with aspect ratios (Ar) of 1 to 5 and dimensionless corner radius (r/D) from 0 to 0.5 were numerically investigated at Reynolds numbers (Re) of 30 to 150 using ANSYS Fluent. Additionally, aerodynamic and flow quantities around the cylinder across various Ar and corner radius were predicted using different machine learning models trained on data from these numerical simulations. Results revealed that the critical Reynolds number increases with rising Ar. For square cylinders, the critical Re increases from r/D = 0 to 0.125, remains constant from 0.125 to 0.375, and decreases from 0.375 to 0.5. For rectangular cylinders (Ar = 2 to 5), the critical Re increases across all Ar as the corner radius rises from 0 to 0.25, with a stronger effect in the 0–0.25 range compared to 0.25–0.5. Two distinct mean flow patterns identified across the Re, Ar, and r/D ranges. In the first flow pattern, the flow remains attached to the cylinder surface and separates from the rear corners, with no vortices forming on the lateral surfaces. In the second pattern, separation occurs from the front corners, generating vortices on the lateral surfaces accompanied by reattachment points. The first pattern dominates rounded-corner cylinders across all examined Re and corner radius, while the second flow pattern appears only in sharp-cornered cylinders at various Ar and Re = 100–150. Instantaneous flow analysis showed that vortex shedding strength behind square cylinders increases with corner radius. However, for rectangular cylinders at fixed Ar, increasing the corner radius yields no significant change in rear vortex strength. Across all corner radius, rear vortex strength decreases with increasing Ar. This study examined variations in aerodynamic coefficients such as fluctuating lift and drag, mean drag (pressure and friction), Strouhal number (St), and fluctuating normal/shear stresses which depend on flow regime, Re, and Ar. Fluctuating lift/drag coefficients and stresses correlate directly with rear vortex shedding strength and decrease with rising Ar. For example, at Re = 150 for sharp-cornered rectangular cylinders, increasing Ar from 1 to 5 reduces fluctuating lift coefficient by 68%, mean drag coefficient by 21%, and St by 16%. For rounded corners (r/D = 0.25), the reductions are 73%, 17%, and 21%, respectively. At Re = 150, square cylinder mean drag decreases by 10% as r/D rises from 0 to 0.25 but increases by 4% from 0.25 to 0.5. In rectangular cylinders across all Ar, mean drag decreases continuously with increasing corner radius. Nusselt number (Nu) increases with Re across all corner radius and Ar. In both steady and unsteady Re regimes, Nu decreases with Ar but increases with corner radius. For instance, at Re = 150, rectangular cylinders show Nu reductions of 43%, 48%, and 51% as Ar rises from 1 to 5 for r/D = 0, 0.25, and 0.5, respectively. The results of nineteen machine learning models showed that Kernel based models like GPR, SVR-RBF, TPS and MLP model had the best performance in predicting all quantities.
