The recent accidents of the Fukushima nuclear power plants were of great concern worldwide. In a nuclear reactor severe accident involving core melt and reactor pressure vessel failure, it is important to provide an accident management strategy that would allow the molten core material to cool down, stabilize thermally and bring the core debris to a coolable state. One countermeasure for the molten corium ejected from the reactor vessel is to retain the core melt on a so-called core catcher as a newly engineered passive corium cooling system residing on the reactor cavity floor.
The retained core melt is cooled by natural circulation flow of water coolant through the inclined cooling channel underneath the core catcher as well as the water pool overlaid on the melt layer. Two-phase flow boiling with downward-facing heated wall in the inclined cooling channel of the core catcher has drawn a special attention because this orientation of heated wall may cause boiling crisis at lower heat flux than that of a vertical or upward-facing heated wall. Although numerous studies have been carried out in experiments and modeling, flow boiling with downward facing heated surface has not been clearly understood due to its complex nature of the problem.
Heat flux partitioning model such as ANSYS-CFX''s wall boiling model calculates the individual heat flux components independently and can be used to predict the overall wall heat flux. However, these models require sub-models for the parameters on the physical heat transfer processes involved. In order to investigate boiling behavior in this inclined cooling channel with downward-facing heated wall and in particular to get some insight for developing a wall boiling model in thermal analysis of the core catcher, a lab-scale experiment was carried out with 1.2 m long rectangular channel, inclined by 10°~ 30°from the horizontal plane, 0.1 m x 0.1 m of channel cross section. The size of the heated wall was 0.06 m wide, 0.75 m long and the heat flux was provided by the Joule heating of the wall using DC electric current. The tests were conducted with near-saturated water at atmospheric pressure. The heat flux was varied in the ranges of 60 ~ 200 ㎾/㎡ for the mass flux of 100 ~ 300 ㎏/㎡s. Ten thin thermocouples of K-type were attached on the back of the heated steel plate at 5 points with a proper electrical insulation for the wall temperature measurement. These wall temperatures were measured to obtain the local heat transfer coefficients.
High-speed video images showed that bubbles were sliding, continuing to grow, and combining with small bubbles growing at their nucleation sites in the downstream. Then the large bubbles coalesced when the bubbles grew too large to have a space between them. Finally, an elongated slug bubble formed nearly covering the heated wall and liquid film under the elongated slug bubble began to evaporate on the heated wall.
The wall superheats were 5 ~ 18℃ for 60 ~ 200 ㎾/㎡ heat flux, 200 ㎏/㎡s mass flux, and 10°inclined angle. The f1ow boiling heat transfer coefficients were obtained by dividing the heat flux by the difference between the wall temperature and the bulk temperature of water. The measured flow boiling heat transfer coefficients for the range of test parameters were 5,000 ~ 12,000 W/㎡K.
To account for the liquid film evaporation in the RPI wall boiling model in ANSYS CFX, the amount of liquid film evaporation was converted to an equivalent nucleate site density. The total nucleate site density consisted of two contributions: typical nucleate sites in liquid region and converted nucleate sites from liquid film evaporation in vapor slug region. With this improved model, numerical analysis using ANSYS CFX code was performed and the predicted wall superheat agreed well with the experimental data, while the original RPI model largely overpredicted the wall superheat.
To validate the model predictions, another numerical analysis was conducted for the KAERI flow boiling experimental data. The model slightly underpredicted the wall superheat of the experimental data. Despite of the limited information of the KAERI experiment, the prediction of wall superheat by the improved model agreed somewhat well with the experimental data.