Abstract
This paper investigates the feasibility of performing reactor physics calculations using a hybrid neutron transport methodology. In the presented implementation, the interface current method is used as the deterministic framework, for which the necessary probabilities appearing in the method are estimated in advance using the probabilistic Monte Carlo code Serpent2. Two grids are used: one fine grid for estimating the scalar neutron flux and a coarse grid for computing
the neutron currents on this grid. Once the probabilities have been estimated, the solutions on both grids are computed deterministically and simultaneously. The main advantage of this implementation lies with the fact that the modelling of possible complex geometries is left to the Monte Carlo solver. In addition, since only within cells probabilities are required, the estimation of such probabilities represents a task having an acceptable computing cost. Several
two-dimensional test cases were developed for benchmarking the framework in two energy groups. For single assembly calculations in an infinite lattice, the deviation of the dominant eigenvalue is smaller than typically 50 pcm. Concerning the spatial distribution of the flux, some acceptable agreement was also obtained, with relative deviations generally smaller than 6%. In some other cases considering a checkerboard pattern of fuel assemblies, though, higher
discrepancies were noticed. It is believed that such discrepancies could be alleviated by implementing a finer resolution in space, angle and energy of the framework. This feasibility study demonstrates the viability of the proposed computational route.
the neutron currents on this grid. Once the probabilities have been estimated, the solutions on both grids are computed deterministically and simultaneously. The main advantage of this implementation lies with the fact that the modelling of possible complex geometries is left to the Monte Carlo solver. In addition, since only within cells probabilities are required, the estimation of such probabilities represents a task having an acceptable computing cost. Several
two-dimensional test cases were developed for benchmarking the framework in two energy groups. For single assembly calculations in an infinite lattice, the deviation of the dominant eigenvalue is smaller than typically 50 pcm. Concerning the spatial distribution of the flux, some acceptable agreement was also obtained, with relative deviations generally smaller than 6%. In some other cases considering a checkerboard pattern of fuel assemblies, though, higher
discrepancies were noticed. It is believed that such discrepancies could be alleviated by implementing a finer resolution in space, angle and energy of the framework. This feasibility study demonstrates the viability of the proposed computational route.
Original language | English |
---|---|
Title of host publication | International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, M and C 2019 |
Publisher | American Nuclear Society (ANS) |
Pages | 2166-2177 |
Number of pages | 12 |
ISBN (Print) | 978-0-89448-769-9 |
Publication status | Published - 2019 |
MoE publication type | A4 Article in a conference publication |
Event | International Conference on Mathematics and Computational Methods applied to Nuclear Science and Engineering - Portland, United States Duration: 25 Aug 2019 → 29 Aug 2019 |
Conference
Conference | International Conference on Mathematics and Computational Methods applied to Nuclear Science and Engineering |
---|---|
Country/Territory | United States |
City | Portland |
Period | 25/08/19 → 29/08/19 |
Keywords
- nuclear reactor calculations
- neutron transport
- hybrid methods
- deterministic methods