Numerical analysis of counter-current flow limitation in a horizontal hot-leg with elbow using RELAP5 and VOF–RANS models

Abstract

During the reflux-cooling phase that follows certain small-break loss-of-coolant accidents (SBLOCAs) in pressurized water reactors (PWRs), the condensate formed in the steam generators must descend through the hot-leg to re-flood the core. This downward liquid stream can be throttled by the counter-current steam rising from the vessel, a phenomenon known as counter-current flow limitation (CCFL). A reliable CCFL prediction is therefore pivotal for estimating the passive cooling capability of the primary circuit and, in turn, for judging the safety margin in SBLOCA scenarios. CCFL is quantified here for the COLLIDER facility (190 mm ID) using two complementary strategies: (i) a three-dimensional VOF–RANS model with variable-density ($\rho$-var) turbulence formulation in OpenFOAM  v2206, and (ii) the one-dimensional system code RELAP5-Mod3 with a linear Wallis-type flooding correlation. Four operating regimes are examined in the range $J_f^{\ast 0.5}=0.10\text{–}0.30$. Results show that the VOF–RANS model reproduces the so-called elevated CCFL — i.e. a controlled overshoot of the Wallis line before full blockage — with errors below 10 % in both pressure drop and blockage onset. In contrast, RELAP5 anticipates blockage by up to 25 %, confirming its conservative bias. Parametric studies reveal that the $\rho$-var formulation lowers the excess interfacial drag by roughly 40 % relative to the incompressible variant, and that mesh refinements finer than 10 mm produce marginal changes in global outcomes. Two practical guidelines emerge: (a) the 1-D approach is adequate for $J_f^{\ast 0.5}<0.15$; (b) for $J_f^{\ast 0.5}>0.20$ or for geometries with pronounced bends, an interface-capturing CFD model is essential to avoid overly conservative blockage estimates. These findings provide a clear basis for selecting and calibrating numerical tools in full-scale nuclear-plant safety assessments.