Test Article
1/5-scale Joby 2017 eVTOL propeller with five blades and diameter D = 0.6096 m. The experimental pitch setting is 16 degrees at 75% span.
Rotorcraft Aeroacoustics
Multi-fidelity analysis of a five-blade eVTOL propeller in hover, edgewise flight, and ground effect, connecting rotor loading, wake redirection, acoustic integration, and noise prediction.
Operating Cases
DDES focuses on 4000 RPM hover and 10 m/s edgewise flight, from the out-of-ground-effect baseline to a 2R rigid-ground case. CHARM then sweeps RPM and clearances of 2R, 1R, and 0.5R.
Validation Data
The simulations are compared with Virginia Tech/NASA-ULI measurements, including thrust, torque, far-field spectra, OASPL, directivity maps, and beamforming trends from a 251-channel microphone array.
Numerical Setup
The high-fidelity simulations use OpenFOAM v2412 with the compressible unsteady PIMPLE solver rhoPimpleFoam. A URANS solution is first developed to accelerate startup, then the flow is advanced with Spalart-Allmaras DDES using improved boundary-layer shielding to prevent grid-induced separation.
The CFD domain is a 25D cube with non-reflecting far-field boundaries. Ground-effect cases add a viscous wall below the rotor, including prism layers and a refined near-ground region so the outwash and reflected pressure waves can develop before acoustic extraction.
Grid and Motion
- Propeller-only geometry, without hub or fairing, to preserve blade-mesh quality.
- Pointwise prism-layer blade mesh coupled to snappyHexMesh with AMI.
- Target blade-wall resolution near y+ = 20 with roughly 40 prism layers.
- Rotating near-blade refinement: D/410 cell size.
- Wake refinement: D/205 near the rotor and D/102 farther downstream.
- Hover mesh: about 45 million cells; ground-effect mesh: about 100 million cells.
DDES Time Marching
The DDES runs use a 0.25 degree azimuthal time step at 4000 RPM, corresponding to 1e-5 seconds. Acoustic data are collected after the unsteady wake is established.
CHARM Role
CDI-CHARM provides a fast free-vortex/panel-method path for thrust, loading, tonal acoustics, and clearance sweeps. Its actuator-line model is useful for trends but does not resolve broadband noise.
Compute Scale
The NASA Pleiades runs used 500 processors. A hover DDES revolution took about 3.5 hours, and roughly 5.5 hours per revolution when collecting acoustic-surface data.
DDES wake and pressure-field structure
Flow Physics
Ground proximity redirects the wake from downwash into radial outwash along the ground plane. That changes blade loading, shifts thrust from the tip toward the mid-span, and strengthens rotor-wake interactions that can feed both tonal and broadband noise.
In edgewise flight, the mesh refinement is rotated using the predicted wake-skew angle, about 43 degrees for the studied case, so the convected wake exits through acoustic-surface end regions rather than cutting through the sidewalls.
Acoustic Post-Processing
Far-field noise is computed in PSU-WOPWOP with the Farassat 1A formulation of the FW-H equation. The study compares impermeable rotor-surface predictions with permeable FW-H surfaces that enclose the rotor wake and part of the near-ground acoustic field.
FW-H surface data are collected for eight propeller rotations at 50 kHz, producing 6000 samples for the 4000 RPM case. OASPL comparisons focus on 1 BPF to 7 BPF, from 333 Hz to 2331 Hz, where the numerical and experimental comparisons are most reliable.
Surface Strategy
- Impermeable surfaces use the propeller surface and need the Method of Images for ground reflections.
- Standard permeable surfaces capture wake-related noise but are sensitive to placement and end-cap treatment.
- The preferred larger permeable surface reduces wake contamination and improves broadband/directivity agreement.
- The ground-effect surface uses an upside-down T shape to enclose the near-ground reflection zone.
- Applying the Method of Images to that T surface would double count reflected waves already captured by the FW-H data.
Vorticity and reflected-wave interaction
Ground-Effect Results
For hover, in-ground-effect operation increases SPL beneath the rotor by up to about 5 dB at smaller clearances, while levels above the rotor can decrease through destructive interference. DDES also shows broadband growth near the ground that CHARM cannot reproduce.
For edgewise flight, the response is more directional and less monotonic with clearance. The BPF level rises when moving from OGE to IGE, but the strongest radiated levels can occur at intermediate clearances rather than at the closest ground position.
Hover Trend
Ground reflections and wake recirculation raise noise below the rotor. At 1R and 0.5R in CHARM, BPF increases are about 5 dB in the region between rotor and ground.
Edgewise Trend
Freestream alignment, wake skew, and blade-vortex interaction make edgewise acoustics strongly directional, with IGE producing roughly 2-5 dB OASPL increases in the DDES comparisons.
Method Insight
CHARM captures the major tonal and loading trends at low cost, while DDES/permeable FW-H surfaces are needed for broadband content and detailed reflected-wave behavior.
Key Takeaways
- The validated setup connects performance, wake physics, acoustic spectra, directivity, and beamforming before applying the model to ground-effect cases.
- Ground effect changes rotor loading by moving lift production from the tip toward the mid-span and redirecting the wake into outwash.
- Hover noise increases most clearly below the rotor, while edgewise noise depends strongly on azimuth, BVI, and clearance.
- CDI-CHARM is valuable for rapid clearance studies, but high-fidelity DDES is required for broadband noise and reflected-wave details.
- The upside-down T permeable FW-H surface captures ground-reflected waves directly and avoids the double-counting risk of combining permeable surfaces with MOI.