High-Performance Computing–Based Simulation of Broadband Acoustic Scattering in Complex Structural Environments

Abstract

The broadband acoustic scattering analysis, in complicated structural settings, is an urgent need in the broad spectrum of engineering usage, comprising the architecture acoustics, underwater-based sensory systems, aerospace sound prediction, and the urban propagation of sound. The true numerical analyses of broadband wave-structure interactions in realistic geometries continue to be difficult as a result of high-frequency content, large-scale-computational domains and complicated boundary conditions. This paper proposes a high-performance computing (HPC)-based simulation model to solve these problems, which is efficient and scalable in analyzing the broadband acoustic scattering of complicated structural setting. It is proposed that the frequency-domain formulation based on a finite element of the proposed methodology will be applied together with parallel domain decomposition and optimized frequency-sweeps on distributed-memory computing architecture. Scalable Krylov subspace solvers and communication efficient parallelization are used to allow the solution of large sparse systems that are a result of high resolution discretization. The frequency sweeps are performed in a systematic manner and the solver configurations are reused to minimize the amount of computation required. Canonical benchmark problems are done numerically to ensure accuracy and then complex structural environment-simulations are done containing irregular geometry and multi-scattering surfaces. The findings not only show that the findings are very consistent with reference solutions over a large frequency range, but also indicates the high retention of a frequency-dependent scattering phenomenon, such as interference patterns and localized resonances. Extensive performance measurements indicate total performance scales nearly linearly with both strong and weak scaling with hundreds of processor cores with no issue and, with weak scaling, with millions of degrees of freedom. The findings substantiate that the suggested HPC-based framework is an accurate, reliable and computationally efficient tool on large-scale analysis of acoustic scattering of broadbands thus ideal in high-end engineering and industrial application.