This paper describes the development of a data base and associated interpolation tool used to perform the validation of FEA thermo-mechanical models designed to verify the structural integrity of a self-pressurized modular Reactor Pressure Vessel (RPV) and its nozzles under service transient thermal loads. The main goal is to assess the element’s size and time steps that provide a confidence level on the obtained solutions.
The validation process implies the definition of the geometry under study, its material’s properties, thermal load conditions and type of mesh element. With all this information, the program gives the user a set of curves for maximum time steps vs. temperature change rates for each typical thickness section in the modeled geometry for a chosen element size. Any point located below those curves assures a solution underneath a user specified allowable error (e.g.: 5%).
All calculations are processed using dimensionless variables in order to create a universal data base enabling the analysis of many different situations of geometries, materials and loads. To improve performance, an Artificial Neural Network algorithm was developed. The resulting application significantly reduces the convergence study time and efforts.