Assessment of TRACE code in predicting multi-dimensional two-phase recirculating flow

• Evaluated TRACE (V5p9) using RPI air–water chimney effect test data. • TRACE captures global two-phase flow structures under various conditions. • Good void fraction predictions achieved for tests without internal rods. • Higher errors found in rod tests due to overpredicted suppression effects. • Lack of intercell shear and K-factor modeling may cause significant errors. This study evaluates the capability of the TRACE (V5p9) system code to predict multi-dimensional two-phase recirculating flow using experimental data obtained from the Rensselaer Polytechnic Institute air–water test facility. The test section simulates a two-dimensional slice of a reactor core and includes configurations both with and without internal rod structures. A total of thirty-six tests simulating chimney effects under various inlet flow conditions were used for assessment, with local void fraction distributions at three vertical levels serving as the primary validation metric. A comprehensive refinement study of nodalization schemes, timestep sizes, and spatial discretization methods was conducted to determine optimal numerical settings. Compared to experimental data, TRACE successfully captured global flow structures, including recirculation patterns and flow regime distributions. For configurations without internal rods, TRACE provided reasonably good void fraction predictions, particularly under low flow rate and equal-split inlet conditions, with mean absolute errors (MAEs) ranging from 3.3 % to 34.2 %. These discrepancies were primarily due to the overestimations of recirculation, attributed to the absence of intercell viscous shear and turbulent diffusion modeling in TRACE. For configurations with internal rods, prediction errors were higher (MAEs of 12.2 %–50.8 %), and void fractions at the upper level were consistently underestimated. These discrepancies are likely caused by overestimation of rod-induced suppression effects, stemming from the simplified treatment of rod bundle structures using equivalent K-factors. Future work will focus on modeling intercell viscous shear and turbulent effect in TRACE and validating the code using diabatic steam-water experiments. These efforts aim to enhance TRACE’s predictive capability for small modular reactor safety analysis.