Development of the control assembly pattern and dynamic analysis of the generation IV large gas-cooled fast reactor (GFR)

During the past ten years, different independent factors, such as the rapidly increasing worldwide demand in energy, societal concerns about greenhouse gas emissions, and the high and volatile prices for fossil fuels, have contributed to the renewed interest in nuclear technology. It is in this context that the Generation IV international forum (GIF) launched the initiative, in 2000, to collaborate on the research and development (R&D) efforts needed for the next generation, i.e. Generation IV, of nuclear reactors. These advanced systems will be ideally deployed beyond the year 2030, following the Generation III or III+ nuclear power plants, which are mainly based on light water technology and are currently entering deployment. A particular goal set for Generation IV systems is closure of the nuclear fuel cycle. Thus, apart from improvements in safety, they are expected to offer a better utilization of natural resources, as also a minimization of long-lived radioactive wastes. Among the systems selected by the GIF, the Gas-cooled Fast Reactor (GFR) is a highly innovative system with advanced fuel geometry and materials (fuel pellets of mixed uranium-plutonium carbide within a plate-type, honeycomb structure made of SiC). It is in the context of the large, 2400 MWth reference GFR design that the present doctoral research has been conducted, the principal aim having been to develop and qualify the control assembly (CA) pattern and corresponding CA implementation scheme for this system. The work has been carried out in three successive and complementary phases: (1) validation of the neutronics tools, (2) the CA pattern development and related static analysis, and (3) dynamic core behavior studies for hypothetical CA driven transients. The deterministic code system ERANOS and its associated nuclear data libraries for fast reactors were developed and validated for the previous generation of sodium-cooled reactors. The validation of ERANOS for GFR applications was, therefore, the first task to be realized in the present research. This has entailed a systematic reanalysis of the GFR-relevant, integral data generated at PSI during the GCFR-PROTEUS experimental program of the 1970's. Thus, during the first phase of the thesis, the reference PROTEUS test lattice from these experiments has been analyzed with ERANOS-2.0 and its associated, adjusted nuclear data library ERALIB1, in order to derive a reference computational scheme to be used later for the GFR analysis. Additionally, benchmark calculations were performed with the Monte Carlo code MCNPX, allowing one to both check the deterministic results and to analyze the sensitivity to different modern data libraries. It has been found that, for the main reaction rate ratios, the new analysis of the GCFR-PROTEUS reference lattice generally yields good agreement – within 1σ measurement uncertainty – with experimental values and with the Monte Carlo simulations. As shown by the analysis, the predictions were in somewhat better agreement in the case of the adjusted ERALIB1 library. The applicability of ERANOS-2.0/ERALIB1 as the reference neutronics tool for the GFR analysis could thus be demonstrated. Furthermore, neutronics aspects related to the novel features of the GFR, for which new experimental investigations are needed, were highlighted. In the second phase of the research, the CA pattern was developed for the GFR, based on iterative neutronics and thermal-hydraulics calculations, 2D and 3D neutronics models for the reactor core having first been set up using the reference ERANOS-2.0/ERALIB1 computational scheme. For the thermal-hydraulics analysis, the CEA code COPERNIC was used. This design work was followed by the study of an appropriate CA implementation scheme (number of CAs and corresponding positions within the core). Detailed neutronics studies revealed the existence of large CA interaction effects, so-called shadowing/anti-shadowing effects leading to an amplification/reduction of the CA worth. The interactions between the absorber pins within a CA, and between the CAs themselves, were investigated in detail, with the goal to optimize the CA efficiency, in terms of the absorber fraction and minimization of the associated heterogeneity effects. The proposed CA pattern consists of 54 absorber pins placed in a triangular lattice. Each absorber pin is a stainless-steel tube filled with highly enriched 10B boron carbide pellets. As a result of the detailed investigations, the absorber pin diameter could be chosen such as to minimize the pin-to-pin influence within the assembly. In particular, a central part of the CA was designed without any absorber pins (zone filled with stagnant helium). A final reduction of the heterogeneity effect (difference between homogeneous and heterogeneous treatments) to 13% was achieved through this feature. Of special importance, the neutronics investigations performed for the reference GFR core ("2004-Core"), especially those related to the CA interactions, have directly contributed to a new core design ("2007-Core"), with the height-to-diameter ratio having been increased to 0.6, compared to 0.3 for the reference core. During the third phase, detailed coupled, 3D neutron-kinetics (NK) and 1D thermal-hydraulics (TH) models were developed for the GFR core, the aim being to arrive at an in-depth understanding of the 3D core behavior during CA driven transients, especially from the viewpoint of spatial effects. The coupled models were developed using the PARCS code for the 3D NK and the TRACE code for the 1D TH modeling. Particular attention was paid to have each individual fuel sub-assembly and CA represented, in order to allow the analysis of local deformations of the 3D distributions of power and safety related parameters, such as the coolant, cladding and fuel temperatures. The validation of the coupled full-core models was performed against reference ERANOS-VARIANT calculations. In particular, the CA worth and reactivity feedbacks were benchmarked, the discrepancies being shown to be relatively low (