A Spherical Multipole Expansion of Acoustic Analogy for Propeller Noise

This paper presents a computationally efficient spherical multipole expansion of Goldstein's acoustic analogy for predicting propeller tonal noise, which decouples source integrals from observer dependence to enable rapid far-field calculations and is validated through simplified lifting-surface and lifting-line formulations that accurately capture dominant radiation characteristics with substantial computational savings.

The Gist

Imagine you are trying to predict the noise a spinning propeller makes. Traditionally, to figure out how loud it is at a specific spot (like a microphone on the ground), engineers have to run a massive, complex calculation that simulates the air rushing over every single inch of the spinning blades. If you want to know the noise at 100 different spots, you have to run that heavy calculation 100 times. It's like trying to predict the weather in 100 different cities by building a new, full-scale climate model for each one individually. It works, but it's incredibly slow and expensive.

This paper introduces a clever new way to do this, which the authors call a "Spherical Multipole Expansion."

Here is the simple breakdown using some everyday analogies:

1. The "Recipe vs. The Meal" Analogy

Think of the propeller's noise generation like cooking a complex dish.

• The Old Way: To serve the dish to 100 guests sitting at different tables, the chef (the computer) cooks the entire meal from scratch 100 times, once for each guest's specific seat.
• The New Way: The chef cooks the dish once and breaks it down into its fundamental ingredients (the "multipole coefficients"). These ingredients are the "source" of the noise.
  • Once the ingredients are prepared, predicting the taste for any guest is easy. You just take the pre-cooked ingredients and apply a simple rule based on where the guest is sitting (the "observer").
  • The Result: You do the hard work once, and then you can instantly calculate the noise for thousands of locations without re-cooking anything. This saves a massive amount of time.

2. The "Orchestra" Analogy

The authors discovered that the noise from a propeller isn't a chaotic mess; it's like an orchestra playing a specific song.

• They found that for most propellers, you don't need to listen to every single instrument to understand the song.
• Just the first two instruments (the "leading multipoles") play 99% of the music.
  • One instrument represents the symmetry of the noise (like the hum of the engine).
  • The other represents the asymmetry (the wobble or unevenness).
• By focusing only on these two "instruments," the computer can predict the sound with amazing accuracy, ignoring the tiny, quiet notes that don't matter much.

3. Two Different "Maps" for Different Blades

The paper also offers two simplified ways to describe the propeller blades, depending on what kind of propeller you have. Think of these as two different maps for navigating a city:

• Map A: The "Lifting Surface" (For flat, wide blades)

  • Imagine a propeller with wide, flat blades (like a helicopter rotor).
  • This method treats the blade like a flat sheet of paper. It's great for calculating noise when the blade is thin and moving slowly through the air. It pays close attention to the thickness of the blade, which is a major source of noise for these types.
  • Analogy: It's like describing a table by its flat surface area.

• Map B: The "Lifting Line" (For long, skinny blades)

  • Imagine a propeller with long, thin, twisted blades (like a high-speed drone or a wind turbine).
  • This method ignores the width of the blade and treats it like a single, thin wire or a "spine" running from the center to the tip.
  • It's perfect for blades that are very long and skinny. It simplifies the math by ignoring the tiny details of the blade's width, focusing only on the lift and drag forces along that central line.
  • Analogy: It's like describing a long, thin noodle by just its length, ignoring its thickness.

Why Does This Matter?

• Speed: Because the hard math is done only once, engineers can test thousands of different propeller designs in the time it used to take to test just one.
• Insight: It helps engineers understand why a propeller is loud. Is it the thickness of the blade? Is it the angle it hits the air? This method separates those factors clearly, like separating the bass from the treble in a song.
• Optimization: It allows for faster design of quieter propellers for drones, electric planes, and helicopters, helping to reduce noise pollution in our cities.

In a nutshell: The authors built a "universal translator" for propeller noise. Instead of re-calculating the physics for every single listener, they calculate the "sound recipe" once and then use a simple formula to tell you how loud it is anywhere you want. It's faster, smarter, and helps us design quieter machines.

Technical Summary

1. Problem Statement

Predicting tonal noise from rotating propellers is a classic challenge in aeroacoustics. Traditional methods, such as the Ffowcs Williams-Hawkings (FWH) equation or Goldstein's acoustic analogy, typically require evaluating surface integrals over the blade geometry for every specific observer location.

• Computational Bottleneck: When predicting noise for a large array of microphones (e.g., for directivity patterns or multi-microphone arrays), these methods require repeated, expensive numerical integrations for each observer point.
• Physical Interpretability: Existing formulations often obscure the relationship between specific aerodynamic mechanisms (lift, drag, thickness) and the resulting acoustic field, particularly regarding the symmetry of the radiation.
• Goal: The authors aim to develop a formulation that decouples source and observer dependencies, allowing for rapid evaluation of the acoustic field at arbitrary locations and providing clear physical insight into the dominant noise generation mechanisms.

2. Methodology

The paper develops a Spherical Multipole Expansion based on Goldstein's acoustic analogy.

Core Formulation

• Acoustic Analogy: The authors start with Goldstein's equation, which expresses acoustic pressure as a surface integral involving aerodynamic force per unit area (dipole) and surface normal velocity (monopole).
• Green's Function Expansion: Instead of integrating the Green's function directly, they expand the free-space Green's function for the Helmholtz equation into a series of spherical harmonics ($Y_{\ell m}$) and spherical Bessel/Hankel functions ($j_\ell, h_\ell$).
• Decoupling: This expansion leads to a separation of variables:
  • Source Coefficients ($A_{\ell m}$): Depend only on blade geometry and aerodynamics (lift, drag, thickness). These are computed once per harmonic.
  • Observer Factors: Depend only on the observer's spherical coordinates ($r_0, \theta_0, \phi_0$) via spherical harmonics and radial propagation factors.
• Resulting Equation: The acoustic pressure at any observer location is a simple summation of the pre-computed coefficients weighted by the observer factors, eliminating the need for repeated surface integrations.

Approximations for Physical Insight

To interpret the physical content of the leading multipole terms, the authors derive two simplified descriptions of the surface integral:

1. Lifting-Surface (LS) Formulation:
   • Assumes small aerodynamic incidence and collapses the blade onto the plane of rotation.
   • Uses thin-airfoil theory to separate contributions into Lift (pressure jump), Drag (pressure sum), and Thickness.
   • Retains chordwise phase effects, making it suitable for large-chord blades and higher frequencies.
2. Lifting-Line (LL) Formulation:
   • Assumes high-aspect-ratio blades where sections are "compact" (chord length $\ll$ radius).
   • Reduces the surface integral to spanwise integrals of compact sectional moments.
   • Naturally handles large incidence angles and induced velocities but neglects chordwise phase effects.

3. Key Contributions

1. Source-Observer Decoupling: The primary innovation is the complete separation of source integrals from observer coordinates. This allows the acoustic field to be evaluated at thousands of observer points with negligible additional cost once the multipole coefficients are computed.
2. Rapid Convergence: The study demonstrates that for hovering subsonic propellers, the multipole expansion converges extremely rapidly. The first two non-zero multipoles (corresponding to the leading symmetric and antisymmetric contributions) capture the vast majority of the acoustic energy and directivity.
3. Direct Radiated Power Calculation: Unlike traditional methods that require integrating pressure over a control sphere to find total power, the multipole formulation provides a direct algebraic expression for radiated power based on the sum of the squared moduli of the coefficients ($|A_{\ell m}|^2$).
4. Physical Decomposition: The simplified LS and LL models explicitly reveal how lift, drag, and thickness contribute to specific multipole orders (e.g., lift drives antisymmetric terms, while thickness and drag drive symmetric terms).

4. Results

The authors validated the method using two baseline propeller configurations (Propeller A: large chord/low incidence; Propeller B: small chord/high incidence) at tip Mach numbers $M_t = 0.2$ and $0.8$.

• Multipole Decay: Tables and figures show that for each harmonic $m$, the coefficients $|A_{\ell m}|$ decay rapidly as the degree $\ell$ increases. For $M_t=0.2$, the first two terms ($\ell=m$ and $\ell=m+1$) dominate. Even at $M_t=0.8$, higher-order terms are negligible for practical directivity reconstruction.
• Directivity Accuracy:
  • The Lifting-Surface (LS) model showed superior accuracy for Propeller A (large chord), correctly capturing chordwise phase effects and directivity shifts at high Mach numbers.
  • The Lifting-Line (LL) model performed exceptionally well for Propeller B (small chord, high incidence), accurately capturing the tilted main lobe and asymmetry.
  • Both approximations matched the "exact" surface integration results closely within their respective regimes of validity.
• Computational Efficiency:
  • A timing benchmark showed the Lifting-Line formulation was ~153 times faster than the exact surface integration.
  • The Lifting-Surface formulation was ~58 times faster.
  • The speedup arises because the simplified models avoid complex 3D geometry preprocessing and reduce integrals to 1D spanwise operations.
• Physical Mechanisms:
  • At low Mach numbers, thickness often dominates the radiated power for low-incidence blades.
  • As Mach number or incidence increases, lift (antisymmetric contribution) becomes increasingly dominant, shifting the directivity lobe away from the rotor plane.

5. Significance

This work provides a powerful, unified framework for propeller noise prediction with three major impacts:

1. Computational Efficiency: It solves the "many observers" problem inherent in modern acoustic testing and optimization, reducing computational costs by orders of magnitude compared to standard FWH implementations.
2. Design Optimization: The ability to calculate radiated power directly from coefficients and the rapid evaluation of directivity make this method ideal for iterative design and noise minimization.
3. Physical Insight: By decomposing the noise into spherical multipoles and linking them to specific aerodynamic mechanisms (lift vs. thickness vs. drag), the method offers a clearer understanding of why noise is generated, aiding in the development of quieter propeller geometries.

The paper concludes that the spherical multipole expansion is a robust, efficient, and physically transparent tool for both near-field and far-field analysis of tonal propeller noise.