The wall pressure distribution of a 3D s-shaped diffuser is affected by the centerline curvature, shape and cross-sections’ area, so achieving a unique solution for the 3D inverse design of a curved duct is a challenging problem in aerodynamic design. All the previous developments on the ball-spine algorithm were limited to 2-D and quasi-3D ducts, in which only the upper and lower lines of the symmetry plane were modified based on the target pressure distribution and the pressure difference is a criterion for the displacement of control points, which is not a suitable criterion in three-dimensional inverse design, due to the effect of area on the pressure contour. In this thesis, two methods for the 3D inverse design of an s-shaped diffuser are presented witch in both methods, a 3D Euler solver is coupled with a geometry correction algorithm. In the first method, the criterion for balls movement is the confused pressure difference, which corrects the duct curvature and the cross-sections in two separate steps. In this method, the effects of the spines (radial, vertical and horizontal) and the grid generation scheme in achieving the unique solution in inverse design were eva‎luated and the results showed that for convergence to a unique solution, it is necessary to use the horizontal spins with the streamline-based grids. The inverse design was eva‎luated by selecting a straight diffuser with semi-elliptic sections as the initial guess and converging to an s-shaped diffuser with an elliptic section as the target geometry. Then, by modifying the target pressure contour on all streamlines on the wall, the high load on the upper and lower lines of symmetry plane was compensated by modifying the load on the side lines of the wall and the corresponding geometry was obtained by 3D inverse design. The aerodynamic performance of the initial and designed s-shaped diffuser was performed by solving the Reynolds-averaged Navier-Stokes equations (RANS) with the k-ω SST turbulence model, which eliminates separation, reduces total pressure drop, and improves outflow distortion in the designed diffuser. The large number of geometry corrections to achieve the target geometry can be considered as a disadvantage of this method. In the second method, using the streamline curvature equation, the dimensionless pressure parameter was defined in which the area effect is eliminated and is a suitable criterion for the displacement of control points. In this method, the 3D diffuser is replaced with several 2D planer ducts, which need to be considered perpendicular to the flow. The dimensionless pressure defined at each control point is a function of the two parameters, centerline curvature and section height (half the distance between the upper and lower lines of a 2D planer ducts). Alternately in each stage of geometry correction, one of the mentioned parameters is considered constant and the other parameter is modified. The proposed method was validated by choosing a straight duct with the desired cross section as an initial guess, witch converges to the target s-shaped duct in the least iteration. A comparison of the two methods shows that the number of geometry corrections in the second method is much less than the first method for convergence. Finally, by applying the second method, the aerodynamic design of an s-shaped diffuser was performed by increasing the height difference between the inlet and outlet of the duct by 20%. In this design, the duct cross-sections were modified by correcting the load pressure on the wall streamline, and by eliminating the secondary flow and separation, the diffuser efficiency was significantly improved by reducing the outlet flow distortion by 69% and increasing the pressure recovery by 30%.

مهدي نيلي احمدآبادي

فرهاد قدك

علي اشرفي زاده، محمود اشرفي زاده، احمدرضا پيشه ور