Document
Computational Support of 9x7 Wind
Tunnel Test of Sonic Boom Models
with Plumes
+ + + James C. Jensen , Marie Denison , Don Durston , and + Susan E. Cliff AIAA-2017-0042 AMS Seminar– Ames Research Center– March 14, 2017 + NASA Ames Research Center, Moffett Field, CA
Outline
Ø Background Ø Methodology Ø Pre-Test CFD Ø Post-Test CFD • Current Methods • Detailed Study • Recent Developments Ø Conclusions
Background
Ø NASA has been working with industry to develop technologies to enable civilian supersonic aircraft to fly over land Ø An ultra low sonic boom aircraft is required to enable FAA certification of commercial supersonic transports Ø Accurately predicting the noise produced by low boom aircraft is critical Lockheed-Martin Low Boom Flight Demonstrator Concept
Background
Ø Pressure signature prediction has two components • Using CFD to calculate the near-field pressure signature • Using atmospheric propagation codes to propagate the signal to the ground Ø Aft signature prediction is a challenge due to the • Boundary layer effects • Interactions of the nozzle plume with shock waves produced by other surfaces of the geometry
Background
Ø A small-scale experiment was performed in the NASA Glenn 1x1 Foot Supersonic Tunnel to examine shock/plume interaction (2014) • Pressure signature taken at 1 nozzle diameter (1.0’’) • 1 model configuration Ø A larger-scale shock/plume test was performed in the NASA Ames 9x7 Foot Supersonic Wind Tunnel that provided data for models that were representative of realistic aft geometries (2016) • Pressure signatures up to 23 nozzle diameters (35’’) with a 1.5” nozzle diameter • 5 model configurations Double Wedge Aft-Swept Deck Nozzle Only Biconvex LaRC 25D 1 2 3 Castner et al, AIAA 2015-1046, Durston et al, AIAA 2017-0041, Cliff et al, AIAA 2016-2035
Background
Ø 160+ experimental runs Ø Two Mach numbers • 1.6 and 2.0 Ø 4 Model altitudes • 8’’, 15’’, 23’’, and 35’’ Ø 3 Roll angles • 0°, 15°, and 30° Ø 2 Model angles of attack • 0° and 3° Ø 4 Nozzle Pressure Ratios • (4, 8, 10, and 14)
Outline
Ø Background Ø Methodology Ø Pre-Test CFD Ø Post-Test CFD • Current Methods • Detailed Study • Recent Developments Ø Conclusions
Methodology: Solver
Launch Ascent and Vehicle Aerodynamics (LAVA) framework Structured Cartesian Structured Curvilinear Unstructured Arbitrary with AMR Polyhedral nd Ø Unstructured Polyhedral Solver (2 Order) • Steady Reynolds-Averaged Navier -Stokes (RANS) • Spalart-Allmaras turbulence model Kiris et al, Aerospace Science and Technology Journal 2016, Computational Framework for Launch, Ascent, and Vehicle Aerodynamics
Methodology: Mesh Generation
Ø The mesh was generated using a near body “core cylinder” mesh and Mach aligned extruded prismatic layers Ø A Mach aligned refinement region exists between the model and the extruded layer boundary Ø Angle of attack is treated by rotating the model relative to the free stream flow vector
Methodology: Boundary Condition
Ø Three types of boundary condition placements were used • Averaged total pressure and temperature surface at the nozzle plenum ( simplified plenum ) • Averaged total pressure and temperature surface at the top of the strut ( full internal flow ) Ø Plenum BC was used for the pre-test external design study and for a fast sweep of post-test CFD Ø Internal flow BCs were used for the pre-test internal flow path design study and for detailed post-test CFD Simplified Plenum Full Internal
Outline
Ø Background Ø Methodology Ø Pre-Test CFD Ø Post-Test CFD • Current Methods • Detailed Study • Recent Developments Ø Conclusions
Pre-Test CFD Objectives
Ø A goal of the experiment was to provide a large database of validation data that can be used to compare to CFD Ø Use CFD simulations to improve the quality of data that would be gathered during the test • Examine the effects of the model nose shock reflecting off the walls of the tunnel and contaminating the plume data • Examine the internal flow of the test rig to ensure the experiment had a well behaved plume
Nose Shock Reflection from Wall
- Reflected Shock Path Ø M=1.6 signature is compromised by nose shock Ø M=2.0 shifts reflected shock aft, providing M=1.6 corrupt region greater extent of contamination-free plume Ø Results lead to an increase in the number of M=2.0 runs during the experiment corrupt region M=2.0
Internal Flow Issues
Ø Original nozzle design was found to have supersonic flow between the orifice plate and the internal rake, additionally, the nozzle throat was not chocked Ø This would lead to poor quality measurements due to: 1. Needing to use a shock correction to the data measured at the internal rake, which was used to determine the nozzle Strut pressure ratio (NPR) 2. Extraneous shocks would Throat propagate from the nozzle into the plume 3. Lack of choking at the throat would lead to a Internal rake Nozzle exit Orifice plate less uniform plume that would be more sensitive to the flow conditions
Internal Flow Path Design
Original Ø The CFD study involved over 30+ RANS runs with different orifice plate designs (including up to 4 staggered plates), nozzle profiles and flow conditions Ø CFD revealed the need to make changes to the orifice plate itself Ø Two designs met the desired requirements: • Placing the choke plate at the nozzle and strut intersection • Proportionally reducing the area of the nozzle exit and Final throat.
Reduced Throat Area (Final) Choke Plate at Strut Base
Pre-Test CFD Summary
Ø The LAVA unstructured solver was successfully used to help improve the experiment: • The simulations of wind tunnel wall effects showed that it was necessary to have more test runs than originally planned at the M=2.0 condition • The internal flow simulations were used to help redesign the nozzle to realize a cleaner plume and subsonic flow in front of the internal Pitot probe
Outline
Ø Background Ø Methodology Ø Pre-Test CFD Ø Post-Test CFD • Current Methods • Detailed Study • Recent Developments Ø Conclusions
Post-Test CFD Current Methods Objectives
Ø Compare CFD results generated with a simplified plenum boundary condition to the wind tunnel test data and assess their validity • All test configurations • Mach number effect • Angle of attack effect • Model altitude effect • Off-track effect • Nozzle pressure ratio effect
Model Density Gradient Comparison
M=2.0 and NPR=8 (Double Wedge NPR=10)
Model Signature Comparison
M=2.0, H=8’’ (Aft-Swept Deck H=15’’), and NPR=8 (Double Wedge NPR=10) Double Wedge Biconvex Nozzle Only LaRC 25D Aft-Swept Deck
Mach Number Comparison
Biconvex Configuration at NPR=8 M=2.0 M=1.6 Ø The Mach number does not have a negative impact on the comparison between the experiment and the CFD Angle of Attack Comparison Aft-Swept Deck at M=2.0, H=15’’, and NPR=8 α =0.3° α =3.3° Ø Changing the angle of attack does not negatively affect the comparison between the experiment and the CFD
Altitude Comparisons
M=2.0 and NPR=8 (Double Wedge NPR=10) Nozzle Only Double Wedge Biconvex Aft-Swept Deck LaRC 25D
Off-Track Comparisons
LaRC 25D at M=2.0 and NPR=8 H=8’’ H=23’' Ø The differences between the experimental and CFD signatures are not worsened by increasing the off-track angle
NPR Comparisons
Nozzle Only at M=2, H=15’’ Aft-Swept Deck at M=2, H=15’’ Ø As the NPR is increased the differences between the CFD and experimental signatures also increase Ø This seems to be caused by the use of the simplified plenum boundary conditions Ø NPR=4 and 8 still match the experiment well but NPR=14 does not
Post-Test CFD Current Methods Summary
Ø The results generated using the plenum boundary condition compared well to the experimental data • Changing the conditions of the simulations (Mach and angle of attack) as well as the location of where the data was extracted (model altitude and off-track angle) did not negatively affect the comparison between CFD and experiment Ø The higher the NPR the worse the comparison between the CFD and experiment
Outline
Ø Background Ø Methodology Ø Pre-Test CFD Ø Post-Test CFD • Current Methods • Detailed Study • Recent Developments Ø Conclusions
Detailed CFD Study with Internal Flow Path
Ø Compare the simplified plenum and full internal CFD results with the experiment to assess their validity Full Internal Simplified Plenum
Pressure Signature Comparisons
Nozzle Only at M=1.6 and H=15’’ NPR=8 NPR=14 Ø At NPR=8 the full flow, plenum, and experiment are all in close agreement but at NPR=14 the plenum boundary condition signature has significant differences as compared to the experiment
Plume Rake Comparisons (M=1.6, NPR=14)
Ø The probe data are approximated from the simulations by applying the Rayleigh Pitot tube normal shock correction at the location of the probes Ø Included corrections for the drooping and AOA of the rake
Density Gradient Comparison (Nozzle Only)
Full Internal Flow Plenum Internal Mach Contour Comparisons (Nozzle and NPR=8) 3 Diameters Downstream Nozzle Exit Near Internal Rake Mach Profiles for Full Internal Flow Mach Profiles for Simplified Plenum
Flow Visualization Comparison (Biconvex)
RBOS Density Gradient Slice From CFD
Detailed CFD Summary and Conclusions
Ø Full flow simulations show significant differences in the plume region relative to the simplified plenum simulations Ø Using the full internal flow generated a better match with the experiment for all of the NPR values Ø Using the plenum BC provides a closer match to the experiment than the simplified plenum BC using the same mesh
Outline
Ø Background Ø Methodology Ø Pre-Test CFD Ø Post-Test CFD • Current Methods • Detailed Study • Recent Developments Ø Conclusions
Radial Boundary Condition Implementation
Ø Extract cut lines from the full flow path solution at the desired location of the inlet BC (e.g., 1 ray at every 45°), and create an ascii file with the averaged radial profile data columns should be in order: radius, static P, static T, velocity normal to the plane of the nozzle Uz , and radial velocity Ur, e.g. in nozzle.dat : # r P T Uz Ur 0.000000e+00 1.934005e+05 2.443122e+02 2.200106e+02 2.892030e+00 6.892365e-05 1.933989e+05 2.443072e+02 2.200447e+02 2.891260e+00 ...
In the input file, specify the BC with the file name, nozzle center point and normal vector: Radial BC specification Extract cut lines from the full flow path solution at the desired location of the inlet BC (e.g., 1 ray at every 45°), and create an ascii file with the averaged radial profile data columns should be in order: radius, static P, static T, velocity normal to the plane of the nozzle Uz , and radial velocity Ur, e.g. in nozzle.dat : # r P T Uz Ur 0.000000e+00 1.934005e+05 2.443122e+02 2.200106e+02 2.892030e+00 6.892365e-05 1.933989e+05 2.443072e+02 2.200447e+02 2.891260e+00 ...
In the input file, specify the BC with the file name, nozzle center point and normal vector: BC_1 { type = nozzle datafiletype = 1 filename = nozzle.dat nozzlecenter = [0.5,0,0] nozzlenormal = [1,0,0] }
Radial BC Signature Comparison (nozzle only)
Ø The radial BC simulation ( red ) agrees well with the full flow path simulation ( blue ) in contrast to the constant stagnation BC based on the internal rake measurements ( cyan )
Radial BC Rake Comparison (nozzle only)
Ø Satisfying agreement between radial BC and full flow path simulations.
Mach Contour Comparison (nozzle only)
Radial BC Radial BC at Nozzle Inlet Simplified Plenum BC Full Flow Path
Summary and Conclusions
Ø The LAVA unstructured solver was successfully used to improve the experiment Ø The results generated using the simplified plenum boundary condition and current meshing practices compared well to the experimental data Ø Using the full internal flow generated a better match with the experiment for all of the NPR values Ø Using the radial BC provides a good comparison with the experiment but with the same mesh as the simplified plenum BC
Future Work
Ø Simulations including the full rake geometry Ø Full span solution computations for plume shear wall reflections Ø Investigate the effects of the tunnel on the signatures Ø Generate solutions using the LAVA curvilinear solver with structured overset meshes Ø Knowledge-based grid refinement studies possibly using LAVA structured code
Acknowledgments
Ø Derek Dalle at Science and Technology Corp. in the Computational Aerosciences Branch at NASA Ames Research Center (ARC) Ø Don Morr at Millennium Engineering and Integration Services Ø Commercial Supersonic Technologies (CST) project under NASA’s Advanced Air Vehicles Program (AAVP) Ø NASA Advanced Supercomputing (NAS) center at ARC
Questions?
Lockheed-Martin Low Boom Flight Demonstrator Concept