Metacomp Technologies participated in the 6th AIAA Propulsion Aerodynamics Workshop held at the 2023 AIAA SciTech Forum and Exposition in National Harbor, MD, USA. The objective of the workshop was to assess the numerical prediction capability of current-generation CFD technology/codes for inlets, diffusers, and nozzles and their propulsion-specific boundary conditions.
The first of two sets of test cases involved the simulation of an inlet to establish the internal shock structure including the terminal shock, resolving boundary layers and their interaction with the shocks, simulating bleed regions and the interaction of the shock waves with the bleed regions, resolving flow about the vortex generators, and resolving the flow at the engine face.
Shown here is the NASA 1507 inlet, which was modeled for the workshop. This inlet is a Mach 3 axisymmetric-spike, mixed-compression inlet that was tested at the NASA Ames supersonic wind tunnel in the late 1960s. The test case is simplified by its mostly axisymmetric geometry and flow features. The freestream conditions were modeled at Mach 3 with a total pressure of 15 psi, total temperature of 616 ˚R and Re/ft of 2 x 106.
Figure 2: Computational flow domain and boundary conditions
A variety of geometry models from low to high-fidelity were simulated using CFD++ :
- M1 :
A 2D axisymmetric and 3D sector with surface patches to define all four bleed zones. - M2 :
A 3D 27˚ sector with surface patches for bleed zones 1 & 2, and bleed hole arrays along with bleed plenums for bleed zones 3 & 4. There are no vortex generators in this geometry; they are modeled using a source term approach. - M3 :
A 3D 27˚ sector with surface patches for bleed zones 1 & 2, and bleed hole arrays along with plenums for bleed zones 3 & 4. Vortex generators are present in the geometry.
Compressible RANS simulations were performed using the one-equation SA model for turbulence closure, adding the rotational-curvature correction (RC) and quadratic constitutive relations (QCR) to the base SA model. The simulations were started without imposing the bleed boundary condition in the porous bleed model for the bleed zones. Once the boundary layers were formed on the centerbody and cowl interior surfaces, the bleed model was activated. The terminal shock eventually stabilizes its position downstream of bleed region 4, and the inlet operates at a supercritical condition. The bleed model pressures were then adjusted so that the bleed rates for each bleed region matched the supercritical flow rates of the wind-tunnel test condition. If the vortex generator source term model was used in a simulation, it was also activated with the bleed model.
CFD predictions of static pressure profiles on the centerbody and cowl, boundary layer rakes and an engine face rake were compared to wind tunnel data, as shown in Figures 3-6.
Figure 3a: Static pressure profile comparisons on the centerbody
Figure 3b: Static pressure profile comparisons on the cowl
Figure 4a: M1 model engine face rake comparisons
Figure 4b: M2 and M3 model engine face rake comparisons
Figure 5a: Throat rake comparisons, centerbody
Figure 5b: Throat rake comparisons, cowl
Table 1 presents the inlet performance compared with test data. The results show the significant improvement in the predictions from low to high fidelity modeling. The highest fidelity M3 model yielded excellent comparisons with test data. The M2 model with the vortex generators modeled via a source term approach also yielded excellent comparisons with data, showing that modeling the VGs via this approach is a viable option.
| Total Bleed Flow Ratio | Engine Face Flow Ratio | Total Pressure Recovery | |
| Test Data | 0.0718 | 0.9460 | 0.8140 |
| M1-2D/Axi | 0.0729 | 0.9218 | 0.7325 |
| M1-3D | 0.0722 | 0.9223 | 0.7364 |
| M1-3D VG Model | 0.0719 | 0.9228 | 0.7816 |
| M3 MIME | 0.0705 | 0.9228 | 0.7836 |
| M2 MIME VG Model | 0.0706 | 0.9229 | 0.7861 |
| M3 PW Coarse | 0.0707 | 0.9166 | 0.7660 |
| M3 PW Medium | 0.0703 | 0.9185 | 0.7750 |
| M3 Pointwise Fine | 0.0705 | 0.9198 | 0.7784 |
Table 1: Inlet performance results from all simulations compared with test data