A Grid-based integral approach for quasilinear problems

For non-homogeneous and nonlinear problems, a major difficulty in applying the Boundary Element Method is the treatment of the volume integrals that arise. A recent proposed method, the grid-based integration method (GIM), uses a 3-D uniform grid to reduce the complexity of volume discretization, i.e., the discretization of the whole domain is avoided. The same grid is also used to accelerate both surface and volume integration. The efficiency of the GIM has been demonstrated on 3-D Poisson problems. In this paper, we report our work on the extension of this technique to quasilinear problems. Numerical results of a 3-D Helmholtz problem and a quasilinear Laplace problem on a multiply-connected domain with Dirichlet boundary conditions are presented. These results are compared with analytic solutions. The performance of the GIM is measured by plotting the L-2-norm error as a function of the overall CPU time and is compared with the auxiliary domain method in the Helmholtz problem.