A Study of Diffusivity in Concert Halls Using Large Scale Acoustic Wave-Based Modeling and Simulation

With advancement and availability of parallel programming, GPU computing, and cloud computing engines, the acoustic wave-based modeling is no longer a rarely used technique for acoustic simulations. Utilizing the computing power of two Amazon's cloud cluster GPU instances (p2.8xlarge with 96GB of graphic memory, 32 vCPUs and 488GB of RAM and p2.16xlarge with 192GB of graphic memory, 64 vCPUs and 732GB of RAM), this study simulates the acoustic environment of large highly rated concert halls in the world based on the surveys carried out by Leo Beranek and Magne Skalevik to investigate the diffusivity of these halls. An acoustic parallel FDTD solver written by Jukaa Saarelma and Lauri Savioja from Aalto University is used for the calculation of sound pressures. This solver is also used as a basis in the author's newly developed Sound Diffusion Coefficient Calculator (SDCC) which calculates the diffusion coefficient of the 3D surface and volumetric diffusers in one-third octave bands. The role of the surfaces and volumes installed in those halls as diffusers along with the effects that they have on the reverberation time of the halls will be studied and analyzed.