A Sparse-Data-Guided Inverse PINN for the Recovery of Thermal-Hydraulic Fields in Supercritical Water Reactor

Abstract Supercritical water reactors (SCWRs) are the generation-IV nuclear reactors having low carbon footprint and capable of providing outstanding thermal efficiency. Precise mapping of wall heat flux is vital for SCWR safety to mitigate localized hotspots and maintain the structural integrity of fuel cladding. But, due to rapid thermophysical property shifts in water's pseudo-super critical zone and limited data availability of the reactor core it becomes challenging to model thermal hydraulics in the pseudo-critical region of water. Again, inverse physics-informed neural networks are known for their capability to reconstruct unknown time dependent profiles when governing physics of any system is known to an extent. Available PINN studies have utilized the dense grid of temporal data, which may not be always available in reality. So, to address these gaps in this study we have established a very limited data guided multi-stage inverse physics-informed neural network (PINN) architecture integrated with SCWR thermal-hydraulic model THRUST that simultaneously recovers transient bulk fluid temperature and transient wall heat flux distributions under the assumption of constant mass flux and a prescribed range of linear pressure drop. A benchmark validated recovery through an analytical temperature field and simulated heat flux profile demonstrated reliable reconstruction. Extensive parametric and ablation studies establish the robustness of the architecture, aiding to the first time PINN integration on the thermal hydraulics of SCWR like system where data sparsity is a real constraint.