Mechanistic CHF model development for subcooled flow boiling in a vertical rectangular channel under low pressure

Research reactors are recommended to be designed with sufficient safety margin for critical heat flux (CHF) in both normal and transient operation, which requires the development of predictive CHF models for both subcooled and saturated flow boiling conditions. While models exist for saturated flow boiling conditions, to date, no mechanistic CHF approach has been developed for subcooled flow boiling in a rectangular channel. In this paper, a new mechanistic model is derived to predict the subcooled flow boiling CHF in a rectangular channel for turbulent upward flow under low pressure. The proposed model is characterized by its consideration of coalesced vapor clot growth due to bubble coalescence not only in the flow direction, but also in the radial direction. In addition, existing liquid sublayer thickness calculation methods were compared to account for the thinning of the liquid sublayer due to the vapor clot growth. The proposed model was verified with a dataset consisting of 126 data points covering the following operational ranges: pressure 1.01-1.13 bar, mass flux 853-15,120 kg/m(2)s, exit quality -0.125 to -0.006, inlet subcooling 8-80 K, channel width 4-22 mm, channel gap 1-6.35 mm, heated length 50-305 mm, and turbulent upward flow. The developed model showed good prediction performance with a 32.47% RMS error, which is the lowest error compared to existing liquid sublayer dryout models developed for circular channels. (C) 2021 Elsevier Ltd. All rights reserved.

Automated summary

A new mechanistic model was developed in this study to predict the subcooled flow boiling CHF in a rectangular channel for turbulent upward flow under low pressure. The model considers the growth of coalesced vapor clot due to bubble coalescence in both the flow direction and radial direction, as well as the thinning of the liquid sublayer due to vapor clot growth. The proposed model showed good prediction performance with the lowest error compared to existing models.