研究RESEARCH研究专题RESEARCH TOPIC
高性能流动声学计算方法及应用High-performance Computational Aeroacoustics
科学问题QUESTION
兼顾声场预测的高精度与复杂环境、大尺度传播问题的计算效率。
Balance high-fidelity acoustic prediction with the efficiency required for large-scale propagation in complex environments.
研究方法METHOD
求解时域声学控制方程，发展能量稳定的高阶通量重构格式，并以高斯波束追踪处理高频传播。
Solve time-domain acoustic equations with energy-stable high-order flux reconstruction and use Gaussian-beam tracing for high-frequency propagation.
验证证据EVIDENCE
通过线化欧拉方程验证数值精度，并分析涡流影响、圆柱散射与海底声线轨迹。
Validate accuracy with the linearized Euler equations and examine vortex effects, cylinder scattering and underwater ray paths.
我们如何研究How we investigate
在高精度方面，专注于发展基于高阶数值格式如通量重构方法的气动声学计算工具。具体来说，通过数值求解时域的声学控制方程如波动方程、线化欧拉方程和欧拉方程等，得到高精度的声场信息，进而系统分析流动与声波的相互耦合问题。当前构造了一种能量稳定的通量重构格式来避免计算中产生数值耗散和色散，并基于一维线化欧拉方程进行了详细的验证和分析。
对于高频的大尺度声传播问题，高效率的求解成为了主要的需求。通过发展适用于具有流动的复杂环境的高斯波束追踪方法，从几何声学的角度高效地求解大尺度的声传播问题，为环境噪声治理、海洋噪声分析等实际问题提供支撑。
当前提出了改进的高斯波束追踪方法来计算流动中声传播的问题，并对二维涡对点声源的声场的影响进行了详细的计算和分析。
面向大规模流动噪声后处理，将 FW-H 积分与多 DCU 并行计算结合；团队公开实现提供 C++、MPI 与加速器支持。
在低马赫数流动中，可压缩性处理也是噪声预测的重要问题。相关成果介绍与多 DCU 求解器一起，补充了从物理建模到声学后处理的研究链条。
For high-fidelity prediction, we develop aeroacoustic solvers based on high-order numerical schemes such as flux reconstruction. Time-domain acoustic governing equations—including the wave, linearized Euler and Euler equations—resolve detailed sound fields and flow–sound coupling. An energy-stable flux-reconstruction scheme is used to control numerical dissipation and dispersion, with detailed verification based on the one-dimensional linearized Euler equations.
For high-frequency sound propagation at large scales, computational efficiency becomes central. Gaussian-beam tracing for complex environments with mean flow provides a geometrical-acoustics route to environmental-noise and ocean-acoustics applications.
An improved Gaussian-beam method has been developed for sound propagation in flow, including detailed analysis of how a two-dimensional vortex pair modifies the field of a point source.
A multi-DCU FW-H solver targets large-scale aeroacoustic postprocessing. The public implementation combines C++, MPI and accelerator support.
Compressibility modelling at low Mach numbers is another focus of noise prediction. This research highlight and the multi-DCU solver connect physical modelling with acoustic postprocessing.
计算与分析图像Computational evidence
