Multimodal nonlinear acoustics in two- and three-dimensional curved ducts

This paper presents an improved, weakly nonlinear multimodal model for acoustics in two- and three-dimensional curved ducts without flow, which extends previous work to include general curvature, torsion, and width variations to analyze effects like wave steepening and acoustic leakage, with potential applications to brass instruments.

The Gist

The Big Picture: Why Do Trumpets Sound "Brassy"?

Imagine you are blowing into a trumpet. When you play softly, the sound is smooth and mellow. But when you blow hard, the sound gets sharp, bright, and "brassy." Why?

It's because the sound waves inside the instrument are getting steeper. Think of a gentle ocean wave rolling onto a beach. If you push it hard enough, the front of the wave catches up to the back, and it crashes into a sharp peak (a shock wave). In a trumpet, this "crashing" happens because the sound is so loud that the air molecules bunch up, creating those sharp, brassy harmonics.

For a long time, scientists could model this "steepening" in straight pipes (like a slide trombone). But most instruments (like a trumpet or a French horn) are curved. They twist and turn. Until now, it was incredibly difficult to calculate how sound behaves when it's both loud (nonlinear) and curved at the same time.

This paper presents a new "super-tool" that finally lets us do the math for curved, loud sound waves.

---

The Problem: The "Traffic Jam" of Math

To understand sound in a pipe, scientists usually break the sound down into different "modes" (like different notes on a guitar string vibrating in different patterns).

• Linear (Quiet) Sound: The modes act like polite drivers in separate lanes. They don't interact much.
• Nonlinear (Loud) Sound: The modes act like a chaotic traffic jam. The loud waves crash into each other, changing the shape of the sound.

The Old Way: Previous models could handle curves, but only for quiet sounds. Or they could handle loud sounds, but only in straight pipes. When they tried to combine them, the math became so messy and heavy that computers would choke on it. It was like trying to solve a Rubik's cube while juggling chainsaws.

The New Way: The authors (Jensen and Brambley) developed a new mathematical framework that is modular and efficient. They realized that the geometry of the pipe (how it bends or twists) and the physics of the sound (how it gets loud) could be separated.

The Core Concept: The "Admittance" (The Pipe's Personality)

The secret sauce of this paper is a concept called Admittance.

Imagine the duct (the pipe) is a person and the sound is a visitor.

• Impedance is how much the person resists the visitor.
• Admittance is how much the person welcomes the visitor.

The authors realized that the "personality" of the pipe (its curves, its twists, its changing width) is fixed. It doesn't care who the visitor is or what song they are singing. The pipe's personality is the same whether you play a soft flute note or a loud trumpet blast.

The Innovation:
Instead of trying to solve the whole problem at once, they first calculate the pipe's Admittance (its personality) from the exit back to the entrance. Once they know the pipe's personality, they can easily predict how any sound source will behave inside it.

This is like knowing a maze's layout perfectly. Once you have the map, you can figure out how a mouse, a ball, or a car will move through it without having to re-draw the maze every time.

The 3D Twist: The "Helix" Effect

The paper goes beyond flat, 2D curves (like a simple bend in a hose) and looks at 3D curves (like a corkscrew or a helix).

• 2D Curve: Imagine a garden hose bent in a circle. The sound waves on the outside of the bend have to travel a longer distance than the waves on the inside. This causes the sound to get out of sync.
• 3D Twist (Torsion): Now imagine twisting that hose like a corkscrew. The sound waves don't just travel longer or shorter; they actually spin as they go.

The authors found that in a twisted pipe, the sound waves get "localized" on the outside of the curve. It's like water swirling in a drain; the energy piles up on one side. This is crucial for understanding how real brass instruments (which are often twisted) actually sound.

Key Findings & Surprises

1. Acoustic Leakage: The authors discovered that even a tiny bit of curvature or a tiny bit of loudness can make sound "leak" through parts of the pipe where it should have been blocked.

   • Analogy: Imagine a sound wave is a car trying to drive through a tunnel that is too narrow for it (it's "cut off"). In a straight pipe, the car stops. But if the pipe bends slightly, or if the car speeds up (nonlinearity), the car suddenly finds a way to squeeze through the wall. The sound escapes where it shouldn't.

2. The "Effective Length" Mystery: When you play a curved instrument louder, the pitch changes slightly.

   • Analogy: Think of a runner on a curved track. If they run slowly, they stick to the inside lane. If they run fast and lean into the turn, they might drift to the outside lane. The outside lane is longer.
   • The paper shows that as sound gets louder, the "effective length" of the pipe changes because the waves are forced to travel along the longer, outer path of the curve. This explains why instruments might go slightly sharp or flat when played loudly.

3. 2D vs. 3D: They compared flat (2D) models to real (3D) models. They found that while the "steepening" (the brassy sound) happens similarly in both, the timing is different. The 3D pipe acts slightly shorter or longer than the 2D prediction, meaning our old flat models were slightly off in their pitch calculations.

Why Does This Matter?

• For Musicians: It helps explain exactly why a trumpet sounds "brassy" and how the pitch shifts when you blow harder.
• For Instrument Makers: It provides a blueprint for designing instruments that stay in tune even when played loudly, or for designing them to have a specific "brassy" character.
• For Engineers: This math isn't just for music. It applies to any system where sound moves through a curved tube, like jet engines, HVAC systems, or even medical devices.

The Bottom Line

The authors have built a universal translator for sound in curved pipes. They took a problem that was previously too messy to solve (loud sound + curved 3D pipes) and broke it down into a clean, efficient system.

They didn't just solve the math; they gave us a new way to "see" the sound. By treating the pipe's shape as a fixed "personality" (Admittance), they can now predict how sound will twist, turn, steepen, and leak in ways we couldn't calculate before. It's a major step forward in understanding the physics of music.

Technical Summary

1. Problem Statement

The paper addresses the challenge of modeling weakly nonlinear acoustics in curved ducts with varying cross-sections, specifically in both two (2D) and three (3D) dimensions.

• Context: This research is motivated by the acoustics of brass musical instruments (e.g., trumpets, trombones), where the "brassy" timbre is caused by nonlinear wave steepening and shock formation within the instrument's bore.
• Limitations of Previous Work:
  • Most existing models assume straight ducts or plane-wave propagation.
  • Previous attempts to model curved ducts (e.g., Félix & Pagneux) were restricted to linear acoustics.
  • A recent 2D nonlinear extension by McTavish & Brambley (2019) suffered from prohibitive algebraic complexity and computational inefficiency because the governing matrices varied with the duct geometry at every point, requiring recalculation at each step.
• Gap: There was no unified, computationally efficient framework capable of handling general curvature, width variation, and torsion (in 3D) while accounting for weak nonlinearity (wave steepening).

2. Methodology

The authors developed a weakly nonlinear multi-modal method that extends the linear multi-modal approach of Félix & Pagneux to include second-order nonlinear effects.

A. Governing Equations

• Derivation: Starting from the inviscid mass and momentum conservation equations for an adiabatic gas, the authors perform a perturbation expansion up to order $O(M^2)$, where $M$ is the Mach number.
• Nonlinearity: The expansion retains linear acoustic terms ($O(M)$) and weakly nonlinear terms ($O(M^2)$), capturing wave steepening and the formation of weak shocks. Entropy perturbations are neglected.
• Coordinate Systems:
  • 2D: Uses a Frenet-Serret frame defined by a centerline $\mathbf{q}(s)$, tangent $\mathbf{t}$, and normal $\mathbf{n}$. The duct has independently varying wall widths $X_+(s)$ and $X_-(s)$.
  • 3D: Extends the frame to include a binormal $\mathbf{b}$ and torsion $\tau$. The duct has a circular cross-section with radius $R(s)$. A phase shift $\theta_0(s)$ is introduced to handle the twisting of the coordinate system (following Germano, 1982).

B. Spatial Modal Decomposition

• Basis Functions: The acoustic pressure and velocity are expanded into a series of straight-duct modes (eigenfunctions of the Helmholtz equation with Neumann boundary conditions).
  • 2D: Cosine modes in the transverse direction.
  • 3D: Bessel functions in the radial direction and trigonometric functions in the azimuthal direction.
• Projection: The governing partial differential equations (PDEs) are projected onto these spatial modes. This converts the PDEs into an infinite set of coupled ordinary differential equations (ODEs) for the modal coefficients.

C. Mathematical Unification and Efficiency

• Key Innovation: Unlike the 2019 model, the authors derived a formulation where the matrices and tensors are invariant along the duct. The geometric variations (curvature $\kappa$, width $X/R$, torsion $\tau$) are factored out as scalar coefficients multiplying these constant matrices.
• Benefit: This allows the matrices to be pre-calculated once, significantly improving numerical efficiency and making the interplay between geometry and nonlinearity easier to analyze.
• Admittance Approach: Instead of solving for pressure and velocity directly, the method solves for the acoustic admittance (ratio of velocity to pressure).
  • A Riccati-style equation is derived for the evolution of the admittance from the outlet to the inlet.
  • Once the admittance is known, a first-order ODE is solved for the pressure from inlet to outlet.
  • This approach decouples the duct's geometric properties from the specific source, allowing for easier analysis of resonances.

D. Numerical Implementation

• Truncation: The infinite system is truncated to a finite number of spatial ($\alpha_{max}$) and temporal ($a_{max}$) modes.
• Numerical Viscosity: To prevent energy accumulation at the highest truncated modes (a common issue in spectral methods for shock formation), a specific numerical viscosity term is added. This term is calibrated to mimic the dissipation of a sawtooth wave.
• Solver: The ODEs are integrated using a 4th-order Runge-Kutta method (adaptive or fixed-step).

3. Key Contributions

1. Unified Framework: A single mathematical framework that handles both 2D and 3D curved ducts, including width variation, curvature, and torsion, under weakly nonlinear conditions.
2. Computational Efficiency: The reformulation of the governing equations to use invariant matrices (separated from geometric scalars) drastically reduces computational cost compared to previous 2D nonlinear models.
3. Generalization of Admittance: Extension of the linear admittance concept to a weakly nonlinear tensor convolution, allowing the acoustic properties of complex ducts to be encoded independently of the source.
4. Validation: The model is rigorously validated against:
   • Analytical solutions for straight ducts (Fubini, Blackstock, Fay).
   • Linear curved duct solutions (Félix & Pagneux).
   • Previous 2D nonlinear results (McTavish & Brambley).
   • Analytical Webster horn solutions.

4. Key Results

The paper presents several numerical examples demonstrating the model's capabilities:

• Wave Steepening: Successfully reproduces the formation of shocks and harmonic generation in straight and curved ducts, matching known analytical solutions.
• Acoustic Leakage: Demonstrates that curvature and nonlinearity can cause "leakage" of acoustic energy. Specifically, an antisymmetric mode (which would be cut-off and reflected in a straight duct) can propagate through a curved duct or a nonlinear straight duct due to mode coupling.
• Torsion Effects: In 3D helical ducts, torsion causes planar sources to become non-planar at the outlet. High-amplitude waves in helical ducts show steepening localized on the outside of the curve.
• 2D vs. 3D Comparison: While wave steepening appears comparable between dimensions, the effective acoustic length of a bend differs. 3D bends appear acoustically "shorter" (waves arrive earlier) than 2D bends.
• Effective Length & Pitch Stability: The study investigates how the effective length of a bend changes with amplitude. A "bend correction factor" is introduced, suggesting that for tight bends, nonlinearity forces waves to travel around the outside of the bend, potentially altering the resonant pitch of an instrument as volume changes.

5. Significance and Future Applications

• Brass Instrument Design: The model provides a powerful tool for understanding the "brassy" sound of brass instruments. It allows for the investigation of how bore curvature and width variation affect harmonic content and pitch stability at different playing volumes.
• Design Optimization: The ability to calculate admittance independently of the source suggests potential for designing instruments with pitch-stable characteristics regardless of dynamic level.
• Future Work: The authors suggest extending the model to include mean flow, acoustic lining (absorbing walls), and more sophisticated numerical solvers (e.g., Magnus-Möbius schemes) to handle singularities more robustly. They also note the potential for modeling open-duct radiation patterns.

In summary, this paper provides a robust, efficient, and unified mathematical tool for analyzing complex acoustic phenomena in curved ducts, bridging the gap between linear geometric acoustics and nonlinear wave propagation, with direct implications for musical instrument physics.