A new parallel 3-D numerical modeling of the electromagnetic fields
- Lawrence Berkeley National Lab., CA (United States)
A new parallel 3D numerical method modeling for electromagnetic (EM) fields in the frequency domain has been developed in this paper. The 3D EM numerical modeling is very important for subsurface EM imaging which is one of the essential tools in the exploration geophysics and environment engineering. The parallel 3D numerical algorithm is described as follows: (1) An accelerating Galerkin finite element method is used to discretize the second order 3D Maxwell`s equation of the magnetic field. (2) A new domain decomposition method is used to decompose the total magnetic stiff matrix into 2{sup n} sub matrices, each one is relative to a subdomain. (3) A local precondition biconjugate iteration is used in parallel to solve the sub-matrix equation in each node of the Massive parallel computer. (4) A modified shared global precondition biconjugate iteration is used in parallel to solve the total magnetic matrix equation. The parallel 3D EM modeling can be used to simulate EM fields from any complex models where (1) the computational domain contains the air-earth interface, (2) the frequency may vary from near dc to that of radar ({approximately} GHz), and (3) arbitrary inhomogeneous electric conductivity {sigma}, magnetic permeability {mu}, and the electric permittivity {epsilon} in the anomalous blocks are involved. In this algorithm, distribution of the jobs in the parallel processing is uniform and the parallel arrangement is done appropriately. The parallel arrangement is done appropriately. The parallel arrangement is done appropriately. The parallel arrangement is done appropriately. The parallel effectiveness is up to 95%. The accuracy of the method is very good.
- DOE Contract Number:
- AC03-76SF00098
- OSTI ID:
- 542997
- Report Number(s):
- CONF-951013--
- Country of Publication:
- United States
- Language:
- English
Similar Records
Incomplete domain decomposition preconditioning for coarse mesh neutron diffusion problems
An incomplete domain decomposition reconditioning method for nonlinear nodal kinetics calculations