Development of two-phase flow model for predicting dryout heat flux in particle packed bed

Abstract

Multiphase flow through porous media has been investigated widely in various research fields such as petroleum, chemical, mechanical and nuclear engineering owing to its numerous potential applications. In particular, it has become important in nuclear engineering after the severe accidents at the Three Mile Island-2 (TMI-2, 1979) and Fukushima (2011) nuclear power plants. In a severe accident progression at a light water reactor (LWR), a porous debris bed formation would be expected when molten core material relocates inside a water pool in reactor cavity due to fuel-coolant interaction (FCI). Therefore, the debris bed coolability is a crucial issue in terms of the severe accident management (SAM) as it directly affects whether the accident is stabilized or not.

The cooling limitation of debris bed is generally described as dryout heat flux (DHF), implying maximum heat flux without dryout. The coolable condition can be obtained only when the amount of evaporated water can be supplied into the porous debris bed. Hence, two-phase flow friction force modeling has been extensively attempted, based on pressure loss measurement data up to date. However, most previous two-phase flow pressure loss experiments were conducted at a restricted void-fraction range up to approximately 0.6, which is far lesser than maximum void fraction of dryout condition when co-current flow is allowed.

In this study, therefore, two-phase flow test in packed particle beds are conducted to gather two-phase pressure loss and void fraction data covering high-void fraction region. The test beds are constructed by packing mm-scale particles, 2- 7 mm in diameter. From the analysis of experimental data, a two-phase flow pattern map, relative permeability terms for both liquid and gas phase and interfacial friction in continuous gas flow region are proposed. At the validation process, the proposed model showed great agreement to the measured two-phase flow data in literature including boiling condition. In addition to this, applying proposed model into DHF prediction was able to predict measured DHF values in literature within 15 % of error, regardless of its flooding condition including top, forced injection from bottom, natural circulated in flow from bottom and lateral water ingression.