Implementation of quadratic axial trial functions in the high-fidelity transport code proteus-MOC

PROTEUS-MOC is a pin-resolved high-fidelity transport code, in which the axial variation of angular flux is represented in terms of orthogonal polynomials. Currently, PROTEUS-MOC employs linear functions and requires relatively fine axial meshes to achieve high accuracy, which increases the number of axial meshes and hence the memory requirement. In this study, aiming to reduce the memory requirement and potentially the computational time by allowing larger axial meshes, we have extended the PROTEUS-MOC transport solution method to quadratic trial functions. Preliminary tests for the performance of quadratic trial functions have been performed using the 3-D C5G7 benchmark problem. Test results showed that for the same axial mesh configuration with relatively large sizes, the quadratic approximation yields about 2 to 5 times more accurate pin powers than the linear approximation, depending on the degree of axial variation of angular fluxes. The quadratic approximation also allows the use of about 3 times coarser axial meshes than the linear approximation for comparable pin power accuracy, which consequently reduces the memory requirement by about 2 times. The memory reduction is not proportional because of the increased number of coefficients in each element from 2 to 3. However, the quadratic approximation did not reduce the computational time as expected because of the deteriorated performance of the pCMFD acceleration scheme due to large axial mesh sizes.