A hierarchical-ACA technique for large-scale acoustic simulations: complex geometries with sound adsorbent materials

Abstract

In this paper a boundary element approach for acoustic simulations based on the hierarchical-matrix format coupled with the adaptive cross approximation (ACA) algorithm and a hierarchical GMRES solver is presented. The cluster tree is generated using preliminary considerations of the prescribed boundary conditions. An improved ACA algorithm, applied, separately, to Neumann, Dirichlet and mixed Robin conditions, is described. Numerical results are presented to show the new approach to be up to 50% faster than conventional ACA approach.