Dry-out characteristics of debris bed under different water injection conditions

A porous debris bed may form through the interaction between melting corium and coolant during a pressurized water reactor severe accident, which continues to emit large amounts of decay heat. The cooling of debris bed is a critical phenomenon and serves as a potential mitigation measure to halt the accident. In this paper, an experimental facility DBC was constructed to study the cooling properties and dry-out heat flux under different water injection conditions and system pressures. The debris bed consisted of stone particles ranging in diameter from 1 similar to 10 mm, with a porosity of 0.37. In the debris bed, a total of thirty-three electrical heating elements, twelve optical fibers equipped with 204 temperature sensors, and thirty-three thermocouples were embedded. Additionally, a computer program MIDEC was developed to investigate the vapor-liquid flow characteristics and dry-out mechanism. The experimental results indicated that the cooling of debris bed can be significantly improved by increasing system pressure. Furthermore, it was observed that the dry-out heat flux under bottom water injection conditions was considerably higher compared to that of top water reflooding. Under top reflooding conditions, the counter-current flow of vapor-liquid impeded the downward flow of coolant, resulting in insufficient coolant supply and leading to a dry-out phenomenon at the bottom region. However, under bottom injection conditions, both vapor and liquid flowed in an upward direction. As a result, the void fraction reached its maximum value at the top region of the debris bed. The distinct vapor-liquid flow characteristics under top and bottom injection conditions resulted in notable discrepancies in dry-out heat flux and dry-out positions. Based on the comparison between the calculated dry-out heat flux (DHF) and experimental results, it was observed that the Reed model accurately predicted the cooling of debris bed under top-flooding conditions when compared with measured data obtained from DBC experiments. However, for very small particle sizes, the Lipinski model would be more suitable and provided better predictions.