An efficient rectilinear Gaussian beam tracing method for sound propagation modelling in a non-turbulent medium

Large-distance sound propagation with high-frequency noise sources, multiple obstacles/geometry with varying acoustic impedance is common in real-life applications. To resolve the acoustic governing equations directly is often computationally costly, especially in three-dimensional space. Methods based on geometric acoustics can be more rapid. However, efforts are still being made to improve the efficiency, robustness, and the capability for complex configurations of such methods. In this paper, an efficient implementation of the rectilinear Gaussian beam tracing method is conducted, which combines rectilinear ray tracing with a proposed efficiency-matched dynamic ray tracing algorithm. A continuous medium stratification method is employed to improve the robustness. Also, a ray compression algorithm is proposed to save computation time. Numerical tests show that computation acceleration up to tenfold is achieved, benefiting rapid estimation of large-distance sound propagation. A standard octree data structure is employed in the code, which accelerates ray tracing in the testing cases with complex geometries. The efficiency and capability of the solver are demonstrated by studying several benchmark problems with varying complexity.