| Summary: | While dealing with a solid-type ceramic breeding blanket for a fusion reactor, it is critical to determine the basic and thermal properties of the functional material in the form of a pebble bed. In the form of pebbles, lithium ceramics serve as the tritium breeder material in the fusion blanket. Effective thermal conductivity (keff) is one of the important thermal properties for the design and useful parameter to determine the performance of the blanket component. Artificial Neural Networks (ANN) are a popular machine learning technique for tuning between input and output parameters. These networks can learn from examples (data set) and apply them when a homogeneous event arises, making them able to work through genuine-time events. Hence, it can save a lot of time and money for doing repetitive experiments and high-end simulations. This will aid in the creation of a huge database on the pebble bed's keff, which will be useful in the design and development of fusion blankets. The findings of simulations and experiments are compared to those predicted by the ANN model for the pebble bed's thermal conductivity.At IPR, a test setup for experiments has been developed using the steady-state and axial heat flow approach. keff of Li2TiO3 has been measured for the pebbles of diameter of 0.8–1.2 mm having packing fraction of ∼62% and using helium environment at different temperatures ranging from 100 °C to 600 °C at constant atmospheric pressure. keff has been compared with pebble bed of stainless steel pebbles of different diameters (1, 2, 3, 1&3, & 2&3 mm) as well. DDPM-DEM model has been used to generate the pebble bed and ANSYS-CFD simulations using FLUENT have been performed to validate the results. The projected values using ANN are within 5% of the results obtained from simulations and experiments. The details of the DDPM-DEM and ANN models, FLUENT simulations, and experimental results will be discussed in this paper.
|
|---|