This paper presents an evaluation of the new radial submesh method of the advanced core simulator SIMULATE-5 (S5) and the effects on pin power reconstruction for an EPR core design. The developed S5 core model is first verified both against results from the predecessor SIMULATE-3 (S3) code as well as to a vendor solution. The latter provides confidence in the developed model while the consistency with the former constitutes an intrinsic verification of the new S5 methodology in terms of performing the critical leakage correction of nuclear data that was previously handled by the lattice code. Thereafter, the impact of various submeshes on the S5 assembly and pin power results is studied, starting from a coarse 1x1 assembly-size mesh typical of conventional nodal codes down to a fine 17x17 submesh structure representative of the lattice full pin-by-pin geometry. Since focus is on the radial submeshing, only 2-D axially-integrated pin powers are investigated and it is found that the various submesh subdivisions give comparable results, with differences typically around ±1%. It is however confirmed that submesh effects increase across- and within fuel assemblies as function of stronger flux heterogeneities and gradients (e.g. core/reflector interfaces and/or control rods). An observation is that submesh "mixture" effects can occur, especially if submeshes include gadolinium bearing fuel rods, and compared to a fine mesh solution, this could produce even larger differences than a solution obtained with a simple coarse assembly size mesh.