Adapting EMAP3D to parallel processing

EMAP3D (Electromechanical Analysis Program in Three Dimensions) is a single -processor (serial) program that uses finite element (FE) methods to solve coupled electromagnetic, thermal and structural problems for high velocity conductors in transient electromagnetic fields. Its primary application has been the simulation of electromagnetic launchers and pulsed rotating machi nes. While EMAP3D has been applied successfully to a wide range of problems , its serial execution time limits the achievement of finer detail in large problems, even on high performance vector machines. The present class of p roduction simulations, involving 100 000 unknowns, can take a week to compl ete on time-sharing machines at computation centers, To reduce simulation t ime, a parallel implementation of the EMAP3D program has been undertaken. W hile vector parallel programming is straightforward and a reasonable approa ch, access to more than 16 vector processors is limited. The availability, low cost, and scalability of Massively Parallel Processin g (MPP) make MPP computing more attractive than vector-parallel processing, The number of R IPP processors on PC Beowulf clusters usually ranges from 8 to 100 and can be as high as several thousand on IBM (SP) and Gray (T3E) systems. We have decoupled the matrix generation and solution components of the FE a lgorithm, thereby allowing us to use any of the new MPP-parallel solvers on the matrix equations, Since the number of zero matrix elements is high for EMAP3D problems, a sparse matrix solver is ideal. Hence, our parallel impl ementation uses the sparse iterative solvers of PETSc (Portable Extensible Toolkit for Scientific Computing). Here we report the performance, scalabil ity, and use of PETSc preconditioners and solver algorithms as a "solution engine" for real EMAP3D simulations and test cases.