Developing 3D neutron transport kernel for heterogeneous structures in an improved method of characteristic (MOC) framework

• Published in: Progress in nuclear energy (New series) 2020-09, Vol.127, p.103442, Article 103442
• Main Authors: Porhemmat, M.H., Hadad, K., Salehi, A.A., Moghadam, A.
• Format: Article
• Language: English
• Subjects: 3-D technology; 3D MOC kernel; Algorithms; Benchmarks; Kernels; Memory management; Modular Ray Tracing; Modular structures; Monte Carlo simulation; Multiplication; Neutrons; Parallel neutron transport; Ray tracing; Simulation; Takeda benchmark; Three dimensional models; Transport equations
• ISSN: 0149-1970; 1878-4224
• DOI: 10.1016/j.pnucene.2020.103442

Given the importance and complexity of the three-dimensional (3D) neutron transport equation solution, in the current research, a new Modular Ray Tracing (MRT) Algorithm and 3D characteristic kernel for heterogeneous structures are presented. Improvement of memory management and cache coherency are achieved to some acceptable level. Also, parallel implementation of transport algorithm utilizing OpenMP, cause significant reduction in runtime. To validate our Algorithm, first, a self-constituted pin cell and a lattice arrangement are modeled and results are compared with Monte-Carlo simulation. Second, the well-known 3D benchmark, Takeda model one and two, are investigated and results compared with the well-known MPACT code. Also, as a reliable trial, the obtained multiplication factors from the forward and adjoint form of MOC kernel are investigated. Meanwhile, for the multiplication factor, we achieved 3.5800E-04 and 3.3924E-04 |Δkeffkeff| the difference with Takeda reference models one and two respectively. Also, the maximum speedup that our parallel algorithm gained for the Takeda benchmark is 5.13. •3D Whole core neutron transport in the MOC framework.•Memory management and reduction in computational cost in solving 3D neutron transport via MOC using new modular ray tracing.•Shared memory parallel Algorithm of the neutron transport equation using the OpenMP interface.•Synchronized implementation of the method of characteristic for neutron transport in forward and adjoint.•The fast and accurate numerical computation which could be extended to whole core calculations.