| Abstrakt: |
Time and precision of the results are the most important factors in any code used for nuclear calculations. Despite the high accuracy of Monte Carlo codes, MCNP and Serpent, in many cases their relatively long computational time causes difficulties in using any of them as the main calculation code. Monte Carlo codes are used mainly to benchmark the results. Ideally, the flux distribution in the lattice would be determined by solving the transport equation in the exact geometry of the lattice using continuous energy cross sections, the way a Monte Carlo code might. But because of time constraints, the calculation scheme within a lattice physics code is intended to reduce the overall computation time without sacrificing too much accuracy. The nuclear deterministic codes have limited precision due to the approximations made to solve the multi-group transport equation. Self-shielding treatment is responsible for the biggest error in any deterministic code; it is an algorithm that produces an average cross-section defined over the complete energy domain of the neutrons in a nuclear reactor. The inaccuracy in deterministic codes is increased dramatically in the resonance period, where a small change in the neutron's energy may lead to a great change in the corresponding microscopic cross section. There are mainly two resonance selfshielding models commonly applied: models based on equivalence and dilution and models based on subgroup approach. The fundamental problem with any self-shielding method is that it treats any isotope as there are no other isotopes with resonance present in the reactor. The most practical way to solve this problem is to use multi-energy groups (50-200) that are chosen in a way that allows us to use all major resonances without self-shielding. This paper offers the best use of a known International Program (DRAGON Code) for nuclear reactor fresh cell calculations by using different methodologies, hypothesis and different Nuclear Data Library, with comparison of other well-known international codes, particularly MCNP5 code and WIMS-D5 Nuclear Data. Similar self-shielding methodologies were tested in other studies. However, all of these studies are concentrated on the low enriched nuclear fuel (3%) which is the standard fuel enrichment of most of nowadays nuclear power reactors. However, there are new designs which propose the use of higher enriched fuel for economical and safety purposes. In higher enriched fuel the effect of resonance interference of uranium isotopes is expected to change. Hence, it is important to test the validity of these self-shielding methodologies and hypothesis in such case. One of these new designs introduces the idea of developing thorium/uranium fueled PWRs. This design has been characterized by its good economics, wide safety margins, minimal waste burden and high proliferation resistance. In such design whole assembly seed and blanket are used, where individual seed and blanket regions each occupy one full size PWR assembly. The seed fuel pin in this design has enrichment close to those used in research reactors (about 20%). In this paper, we selected the fresh seed fuel pin, which is used in thorium/uranium reactors, to be our physical model. Then, we performed cell calculations by solving 172 energy group transport equation using the deterministic DRAGON code. Two types of selfshielding models (equivalence and dilution models and subgroup models) are used. The data libraries used are WIMS-D5 and DRAGON libraries. To obtain the accuracy of the self-shielding treatments, the results are compared with the result obtained from the stochastic MCNP5 code. We also tested the sensitivity of the results to a specific change in selfshielding method implemented, for example the effect of applying Livolant-Jeanpierre Normalization scheme and Rimman Integration improvement on the equivalence and dilution method, and the effect of using Ribbon extended approach on sub-group method. [ABSTRACT FROM AUTHOR] |
