4. METHODS
It is based on the 4th order polynomial nodal expansion method (NEM). As the 4th order NEM is not sensitive to mesh sizes, accurate calculation is possible by the use of coarse meshes of about 20 cm.
The drastic decrease of number of unknowns is a 3-dimensional problem results in very fast computation. Furthermore, it employs newly developed computation algorithm "boundary separated checkerboard sweep method" appropriate to vector computers. This method is very efficient because the speedup factor by vectorization increases, as a scale of problem becomes lager. Speed-up factor compared to the scalar calculation is from 20 to 40 in the case of PWR core calculation. Considering the both effects by the vectorization and the coarse mesh method, total speedup factor is more than 1000 as compared with conventional scalar code with the finite difference method.
The general theory of NEM, the fast computation algorithm, benchmark calculation results and detailed information for usage of this code including input data instruction and sample input data is described in the documentation.
