PRELIMINARY THERMAL-HYDRAULIC ANALYSIS OF THE AHTR FUEL ELEMENT

The Advanced High Temperature Reactor (AHTR) is a molten-salt cooled reactor utilizing TRISO particle based fuel plate with graphite matrix. Its fuel type leads to low heavy metal loading which may challenge the desired cycle length and require increasing the core volume and/or reducing the power density. Therefore, tradeoffs are necessary in the core thermal and neutronic design to provide adequate cooling and moderating capability, while preserving the heavy metal loading. The cooling system must be designed to provide efficient heat removal, reducing the total core volume and the maximum fuel temperature. For preliminary analysis of the thermal performance of the system, a MATLAB model of the single coolant channel and fuel plate was developed. The model provides a steady-state characterization of the coolant temperature, velocity and other physical properties, as well as the temperature distribution within the fuel plate. The power density distribution of the plate, both in the axial and transversal directions, was studied in order to determine realistic operating conditions and to evaluate the effects on the temperature distribution. The MATLAB model was then used to perform sensitivity studies on the main parameters of the assembly design: graphite conductivity change due to irradiation and temperature, sleeve thickness, fuel stripe thickness, TRISO particle packing fraction, coolant channel thickness; also, the combined effect of the coolant gap and fuel stripe thickness variation was considered. A RELAP5-3D model of the coolant channel and fuel plate was developed, in order to validate the MATLAB model and provide the capability for transient simulations. A pipe component was selected for the modeling of the coolant channel and a slab heat structure was selected for the fuel plate. A comparison between the MATLAB and RELAP5 model was performed both with a uniform and cosine axial power density distribution, showing that the difference between the two models is mainly due to the discretization of the power profile in the RELAP5 model. An adapted MATLAB model was developed to evaluate the changes between the continuous and the discretized cosine power profile. A RELAP5 model with increased number of axial intervals was tested and a lower error was obtained. The differences between the two models were considered acceptable and the RELAP5 model will be used for implementation in full core simulations.