Physics-based Inferencing for Characterization of Boundary Conditions on Thermal-fluid System for Irradiation Vehicle

Abstract

This work is an ongoing attempt to develop a method to infer, based on a limited number of point-wise sensor measurements, critical physical distributions of interest in irradiation vehicle. The problem involves accurate characterization of boundary conditions of thermal-fluid. Existing inferencing methods based on the general transformation or data-driven transformation cannot provide efficient reconstruction with few measurements since physics is not included. The work showed that using a physics-based transformation with the appropriate basis for thermal boundary condition, it is possible to reconstruct a steady-state temperature field in a turbulent pipe flow in a 2d axis-symmetric simulation. This work further estimated the inlet mass flow rate if it is unknown. The inferencing technique has to be developed towards more complex turbulent flow phenomena, sensor location and modality optimization, and noisy and imperfect sensor measurements to be applied for the real problem of interest.