System code validation in periodic two-phase flow from low-pressure oscillations

Abstract

Low-pressure two-phase flow instabilities can potentially challenge start-up transients of water-cooled nuclear reactors. Predicting oscillations caused by flow instability is therefore paramount to reactor safety. While many thermal hydraulics system codes are developed as general-purpose tools, their performance is not guaranteed outside their optimized application ranges. The current study therefore performs validations of two thermal hydraulics system analysis codes, the Adaptive SYStem Thermal-hydraulics Version 3 (ASYST VER3) and Reactor Excursion and Leak Analysis Program MOD3.3 (RELAP5/MOD3.3), in reproducing oscillatory two-phase flow induced by low-pressure flashing instability. Benchmark conditions are selected from a novel dataset resolving void fraction with radial resolution and quantifying periodic behavior with ensemble averaging. Focusing on the prediction of transient two-phase phenomena rather than the determination of stability, validations are conducted by simulating a single channel under prescribed periodic boundary conditions. ASYST exhibits a systematic underprediction of void fraction in an adiabatic chimney downstream of a heated section. The prominent causal discrepancy is identified as a lost travelling void wave due to overpredicted condensation. RELAP5 noticeably overpredicts subcooled boiling and underpredicts flashing, which is consistent with existing validations against steady-state separate-effect tests under low pressure. This degraded code performance also suggests that in low-pressure transients prone to flashing instability, the confidence in RELAP5 derived from existing validations against integral-effect tests shall be carefully limited to validated capabilities in integral systems. In general, the current study fills the previous gap of knowledge about the performance of ASYST and RELAP5 in predicting detailed transient two-phase phenomena in low-pressure flashing-induced oscillations. The identified code defects reveal future directions for code improvements eventually towards reliable application of ASYST and RELAP5 under low pressure beyond their original calibration.