Inverse optimization design of heat transfer performance for the reactor pressure vessel insulation structure based on reduced-order model

Based on the above three optimization objectives, a dimensionless comprehensive score on a percentage scale was established to evaluate the overall heat transfer performance. [Results] The POD reconstruction achieved high fidelity, with an average determination coefficient exceeding 0.95 and mean relative deviation (MRD) below 2.87%, while reducing CPU time by 99% compared to CFD. The optimal variable combination (2 mm-16℃-309.5℃) yielded a comprehensive score of 97.4, improving overall heat transfer performance by 59.2% over the original design.

[Conclusions] The POD-based methodology provides an efficient and precise approach for RPV insulation structure optimization, significantly enhancing safety and operational efficiency.