An LLM-assisted theoretical model for aeroengine installation noise scattering

Xun HuangarXiv 2026 · arXiv

This paper presents a novel theoretical approach to a classic problem in aeroacoustics: how noise from an aeroengine duct scatters when it hits nearby aircraft structures like wings or pylons. Beyond the fluid dynamics, the paper serves as a compelling case study for the "human-in-the-loop" use of Large Language Models (LLMs) to perform the grueling symbolic algebra required for theoretical research.

TL;DR

Researchers have developed an exact theoretical model for aeroengine installation noise by "chaining" two classical mathematical solutions through a Weyl–Sommerfeld spectral bridge. By delegating the tedious coordinate transformations and symbolic algebra to an LLM, the authors accelerated the derivation process without sacrificing the rigour required for theoretical fluid mechanics.

What Problem Does This Paper Address?

In aerospace engineering, "installation noise" refers to how engine noise interacts with the airframe. While we understand how noise radiates from a cylindrical engine duct and how waves scatter off a flat plate (like a wing) individually, coupling these two is mathematically difficult.

The engine is best described in cylindrical coordinates, while the wing is best described in Cartesian coordinates. Translating the complex, "spinning" noise modes of a jet engine into a format that a flat-plate model can process usually requires expensive, time-consuming numerical simulations. The authors sought a "closed-form" theoretical model that is faster and provides more physical insight than a "black-box" simulation.

Methods

The authors used a "two-stage" approach based on the Wiener–Hopf technique, a powerful method for solving wave-scattering problems.

  1. Stage 1 (The Duct): They solve the radiation of noise from a semi-infinite cylindrical pipe.
  2. Stage 2 (The Plate): They solve the scattering of that noise by a nearby semi-infinite rigid plate.
  3. The Bridge: The "mathematical crux" is the Weyl–Sommerfeld identity. This identity acts as a translator, expanding the cylindrical harmonics from the engine into a double angular spectrum of plane waves that the Cartesian plate model can "understand."

The LLM Role: The authors utilized an LLM (specifically Kimi by Moonshot AI) as a "mechanical assistant." While the humans provided the core physical insights and the strategy for chaining the models, the LLM handled the "heavy algebra" of the coordinate transformations and assisted in writing the code for numerical verification. Every step performed by the AI was independently validated by the researchers to ensure accuracy.

Datasets

Not discussed in the paper. As a theoretical study, the work relies on analytical derivations and numerical verification against established solvers rather than training on external datasets.

Key Insights and Ideas

  • The Exact Spectral Bridge: The realization that the Weyl–Sommerfeld identity provides an exact (not approximate) link between cylindrical duct modes and Cartesian plate gusts is the paper's primary theoretical contribution.
  • Division of Labor: The paper proposes a productive pattern for AI in research: humans handle the "idea" (physical insight, causality, and logic), while the AI handles the "mechanics" (tedious symbolic manipulation and boilerplate coding).
  • Jet Refraction: The model shows that the presence of a subsonic jet significantly refracts the noise lobes. This shift in angle changes which "interference fringes" the noise hits on the wing, potentially reinforcing or canceling the noise by several decibels.

Key Results

  • High Precision: The spectral bridge reconstructed the incident velocity on the plate with a relative error as low as $10^{-4}$ to $10^{-3}$ compared to finite-difference evaluations.
  • Shadowing and Interference: The model successfully captured the "shadow zone" behind the plate (where noise levels drop by 23–36 dB) and the complex fringe patterns caused by interference between direct and reflected noise.
  • Installation Effects: In one case study, the jet refracted the main noise lobe from $22^\circ$ to $48^\circ$. Because of this shift, the noise hit a "constructive" interference fringe, resulting in a 6.4 dB reinforcement of the noise compared to the engine alone.

Limitations and Future Work

  • One-Way Coupling: The model assumes the plate is far enough away that noise scattered by the wing doesn't bounce back into the engine duct. While reasonable for many setups, an iterative scheme would be needed for very close couplings.
  • Simplified Geometry: The current model uses a semi-infinite, rigid, zero-thickness plate. Real wings have thickness, finite spans, and different surface impedances.
  • Serrated Edges: The authors note that this framework is particularly well-suited for extension to "serrated" edges (often used for noise reduction), which would transform the scalar equations into more complex matrix equations.
  • Instability Waves: The model currently omits Kelvin–Helmholtz instability waves, which are hydrodynamic disturbances that grow in the jet's shear layer but are often saturated by nonlinear effects in real-world scenarios.

Summary updated

Abstract

A theoretical model is developed for the canonical problem of aeroengine fan noise impinging on, and scattering from, a nearby semi-infinite rigid plate. The model chains two Wiener--Hopf problems in different coordinate systems. The first stage solves the duct-radiation problem analytically in axisymmetric cylindrical coordinates. The second stage expands the field incident on the plate into a double angular spectrum of plane waves: the axial spectrum is inherited from the first stage, while the spanwise spectrum is generated by a Weyl--Sommerfeld expansion of the cylindrical harmonics. Each spectral component is then scattered by the plate. The mathematical crux is the coordinate transformation between the axisymmetric duct frame and the Cartesian plate frame. The core physical insight--that the Weyl--Sommerfeld identity provides an exact spectral bridge between these systems--originates entirely with the authors. The heavy algebra of translating between the two spectral representations, together with the associated numerical verification, was carried out with the assistance of a large language model (LLM), under strict human oversight and independent numerical cross-checking. This division of labour enables the LLM to handle tedious symbolic manipulation without compromising rigour. Results are presented for spinning modes radiating from a cylindrical duct with a subsonic jet into a quiescent ambient. The case study demonstrates that, under systematic verification, LLMs can accelerate tedious algebra in theoretical fluid mechanics, contributing a concrete empirical example to the broader discussion on AI's role in theoretical research.
BibTeX
@misc{huang2026llm,
  author = {Xun Huang},
  title = {{An LLM-assisted theoretical model for aeroengine installation noise scattering}},
  year = {2026},
  eprint = {2611.18796},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.18796},
}
Figure