Development and Status of a Finite-Volume, Time-Domain CEM Research Code with Multidisciplinary Applications

A general-purpose finite-volume, time-domain electromagnetic solver is described. The objective of the effort began as a modification to an existing computational fluid dynamics code, Cobalt60, to solve the equations of electromagnetism. The existing framework for handling geometries with unstructured grids and the parallel computing capability made the code conversion convenient and timely. The code implements the solution of the Maxwell "curl" equations. Results for perfectly conducting spherical surface are presented and compare favorably with theory and other electromagnetic solvers. Example results are presented for complex geometries and tests were performed for other benchmark geometries. notable feature (portability) of the present code is its parallel computing capability from high performance platforms to Linux-based clusters. The code also takes advantage of implicit time advance to overcome the typical limitations of explicit methods. Work described includes (i) systematic analysis and implementation of the basic algorithms, boundary conditions and applications; (ii) proof that the code can achieve accurate results in comparison to other methods; and (iii) advantages of implicit time-stepping approach as compared to other finite-difference and finite-volume explicit methods. These claims are substantiated by various numerical simulations performed by the present code and selected alternatives. Open issues are presented and discussed in context of ongoing plans for solution verification and code validation in tandem with continued development. It remains an open question whether codes based on time-domain methods will become computationally efficient to compete with frequencydomain solvers.