FastBEMT

FastBEMT is a compact Blade Element Momentum Theory solver with tonal-noise, broadband-noise, and blade-stress utilities. NumPy/Pandas handle the aerodynamic solution and PyTorch handles the kinematic and acoustic tensor calculations.

Models

• BEMT section loads and integrated thrust, torque, coefficients, and figure of merit
• Farassat 1A compact-source thickness and loading noise
• Brooks-Pope-Marcolini broadband noise components
• Centrifugal and bending stress on the supplied airfoil sections

The maintained package lives under src/FastBEMT. Historical implementations are available through Git history rather than being shipped in the Python package.

Installation

FastBEMT requires Python 3.12 or newer.

uv sync

Notebook and development tools are optional:

uv sync --extra examples
uv sync --extra dev

Geometry contract

Propeller consumes the current PropGen-style geometry mapping. The required entries are:

• airfoils: (sections, points, 2) normalized airfoil coordinates
• r, chord, twist, sweep, and rake: one value per section
• hub_radius, tip_radius, and n_blades

The bundled Data/10x7E.pkl is the reference contract. Radial widths, airfoil areas, boat-tail angles, active hub-to-tip sections, and device tensors are derived once by Propeller.

Aerodynamic analysis

from FastBEMT import (
    BEMT,
    Environment,
    Propeller,
    Simulation,
    load_propeller_geometry,
)

geometry = load_propeller_geometry("Data/10x7E.pkl")
environment = Environment()
simulation = Simulation(
    revolutions=2,
    timesteps_per_revolution=100,
    device="cpu",
)
propeller = Propeller(geometry, environment, simulation)

bemt = BEMT(propeller, rpm=7000, v_inf=0.0)
print(bemt.performance_for())
section_results = bemt.solution_for()

rpm and v_inf may also be one-dimensional arrays; FastBEMT evaluates their Cartesian product. Supply advance ratio J instead of v_inf when that is the natural operating-point input.

Section result names include units or physical meaning where ambiguity would otherwise be likely, for example inflow_angle_deg, section_thrust, relative_velocity, and reynolds_number.

Aeroacoustics

from FastBEMT import BPM, F1A
from FastBEMT.Aeroacoustics import semicircular_observer_array

observers = semicircular_observer_array(radius=2.0, n_points=19)
one_revolution = 60.0 / 7000.0

f1a = F1A(propeller, bemt)
f1a.run(
    observers,
    observer_duration=one_revolution,
    n_observer_times=100,
)

bpm = BPM(
    propeller,
    bemt,
    turbulence_length_scale=0.01,
    turbulence_intensity=5e-4,
)
bpm.run(
    observers,
    observer_duration=one_revolution,
    n_observer_times=100,
)

F1A also accepts the current VPM .pt payload through loadings=. That payload contains a finite (time, blade, section, 4) tensor with source time in channel zero and global-frame force per unit span in the remaining channels, plus embedded r_mid_m and dr_m section arrays.

Layout

• Aerodynamics/BEMT.py: operating-point orchestration and integrated performance
• Aerodynamics/Section.py: one-section aerodynamic solve
• Aeroacoustics/F1A.py: tonal noise
• Aeroacoustics/BPM.py: broadband noise
• Kinematics/Kinematics.py: rotating section frames and derivatives
• Utils/Plotter.py: acoustic maps and stress plots
• Utils/Stress.py: structural calculations

Validation

The maintained automated tests are under tests. The tracked BEMT and stress notebooks retain higher-cost comparison workflows, while obsolete notebooks for the removed all-in-one Propeller.run_aeroacoustics API have been retired.

uv run pytest
uv run ruff check src tests