Influence of the model coefficient changes in the Menters k-ω SST turbulence model on the parameters of the flow around the aerodynamic airfoil.

Abstract

The studies presented in the article relate to the field of aircraft aerodynamics, they are aimed at improving the accuracy of separation flows modeling with using the RANS turbulence model. To conduct the research, the method of mathematical modeling was used with using a package of computational fluid dynamics application programs. Based on the results of numerical modeling, the effect of reducing the constant a1 the RANS turbulence model (k-ω SST of the Mentor) on the parameters of the flow around the airfoil NACA 0012 was estimated. The investigated flow is assumed to be incompressible, threedimensional, the Reynolds number Re = 3.106, the range of the studied angles of attack  = 0…15. In study used hexagonal grid model, the boundary layer is made from the condition of limiting the dimensionless distance to the wall Y+ < 3. To control the quality of the grid model, the independence of the solution from the dimension of the grid was confirmed: the convergence of the lifting force acting on the profile with a systematic decrease in the size of the element was investigated. Validation of the flow in the boundary layer was carried out by comparing the calculated dimensionless velocity profile with the empirical “wall function”. To assess the influence of the value of the constant a1 on the flow parameters, a comparison of the distribution of the pressure coefficient over the profile surface for different values of a1 was carried out. For the value of the angle of attack  = 15, the simulation results using the DES (LES/ Menter’s k-ω SST) turbulence model obtained using the Tornado supercomputer, SUSU are presented. The obtained results confirmed the improvement of the simulation results with a decrease in the value of the a1 coefficient by 30% in the angle range of no more than  = 15. This does not agree with the results presented earlier in the publications of other authors predicting an improvement in results to  = 30. Increasing the value of the a1 coefficient by 30% at  = 15 improves the simulation results slightly.