Flow-induced noise prediction around a cylinder using deep neural network

Minjoon Kim; Imjun Ban; Yumi Lee; Sung-chul Shin

doi:10.7776/ASK.2025.44.2.170

All Issue

2025 Vol.44, Issue 2 Preview Page Next Page

Research Article

Flow-induced noise prediction around a cylinder using deep neural network 심층 신경망을 활용한 실린더 주위의 유동소음 예측

31 March 2025. pp. 170-176

PDF XML

Abstract

Flow-induced noise is generated around submarine appendages due to the surrounding fluid field and can be numerically predicted using Computational Fluid Dynamics (CFD). Predicting flow-induced noise through CFD requires turbulence models based on spatial averaging operations such as Large Eddy Simulation (LES) or Detached Eddy Simulation (DES). However, DES requires significant computational resources due to its high grid resolution and short time steps leading to long overall simulation times. In this study, a Deep Neural Network (DNN) model well-suited for numerical data prediction was employed to predict the flow-induced noise around cylindrical structures. Inlet flow velocity and cylinder geometry were selected as design variables to generate training data. CFD simulations were performed using DES turbulence to calculate the Sound Pressure Level (SPL) at point receivers around the cylinder for each octave band. The design variables, point receiver locations, and octave band center frequencies were used as input features for DNN model to predict the SPLs at each octave band. Through this approach, designers can rapidly estimate flow-induced noise based on the geometry of submarine appendages and flow velocity thereby reducing the time required for design process.

Keywords

Deep Neural Network (DNN)

Deep learning

Flow-induced noise

Submarine appendages

잠수함 부가물은 외부 유동에 의해 소음이 발생하게 되고, Computational Fluid Dynamics(CFD)를 활용하여 수치적인 예측이 가능하다. CFD를 통해 유동소음을 예측할 시, Large Eddy Simulation(LES), 혹은 Detached Eddy Simulation(DES)과 같은 공간 필터링 연산 기반 난류 모델을 사용해야 한다. 하지만 DES는 높은 격자 해상도와 짧은 시간 간격으로 인해 전체 해석 시간에 많은 컴퓨팅 자원이 필요하다. 본 연구에서는 딥러닝 모델 중 수치형 데이터 예측에 적합한 심층 신경망(Deep Neural Network, DNN)을 활용하여 실린더 형상 주위의 유동소음을 예측하고자 한다. 학습 데이터를 생성하기 위해 유속과 실린더 형상을 설계 변수로 선정하고, DES 난류 모델을 활용하여 실린더 주위 수음점에서의 옥타브 밴드별 음압 레벨을 계산하였다. 설계 변수와 수음점 위치, 옥타브 밴드의 중심 주파수를 신경망 모델의 입력 변수로 하여 옥타브 밴드별 음압 레벨(Sound Pressure Level, SPL)을 예측하는 DNN 모델을 개발하였다. 이를 통해 설계자는 설계 과정에서 잠수함 부가물 형상과 유속에 따라 발생하는 유동소음을 신속하게 예측하여 설계에 소요되는 시간을 단축할 수 있다.

키워드

심층 신경망

딥러닝

유동소음

잠수함 부가물

References

J. Ha, G. Ku, J. Cho, C. Cheong, and H. Seol, "Numerical comparative investigation on blade tip vortex cavitation and cavitation noise of underwater propeller with compressible and incompressible flow solvers" (in Korean), J. Acoust. Soc. Kr. 40, 261-269 (2021).

G. Ku, S. Ryu, and C. Cheong, "Numerical investigation into cavitation flow noise of hydrofoil using quadrupole-corrected Ffowcs Williams and Hawkings equation" (in Korean), J. Acoust. Soc. Kr. 37, 263-270 (2018).

M. Wang, J. B. Freund, and S. K. Lele, "Computational prediction of flow-generated sound," Annu. Rev. Fluid. Mech. 38, 483-512 (2006).

10.1146/annurev.fluid.38.050304.092036

B. Joe, S. Hong, and J. Song, "Flow noise analysis of ship pipes using lattice boltzmann method" (in Korean), J. Kor. Soc. Mar. Env. Saf. 29, 512-519 (2023).

10.7837/kosomes.2023.29.5.512

S. Lee and D. Youn, "Analysis of a convolutional neural network for predicting unsteady volume wake flow fields," PoF. 33, 035152 (2021).

10.1063/5.0042768

H. Meng, Z. Xu, J. Yang, B. Liang, and J. Cheng, "Fast prediction of aerodynamic noise induced by the flow around a cylinder based on deep neural network," Chin. Phys. B. 31, 064305 (2022).

10.1088/1674-1056/ac5e98

P. R. Spalart, "Detached-eddy simulation," Annu. Rev. Fluid. Mech. 41, 181-202 (2009).

10.1146/annurev.fluid.010908.165130

Information

Publisher :The Acoustical Society of Korea
Publisher(Ko) :한국음향학회
Journal Title :The Journal of the Acoustical Society of Korea
Journal Title(Ko) :한국음향학회지
Volume : 44
No :2
Pages :170-176
Received Date : 2025-02-23
Revised Date : 2025-03-18
Accepted Date : 2025-03-19
DOI :https://doi.org/10.7776/ASK.2025.44.2.170

The Journal of the Acoustical Society of KoreaISSN:1225-4428(Print) 2287-3775(Online)한국음향학회

All Issue