Computer analyses on loop seal clearing experiment at PWR PACTEL

Abstract

Water seal formation in the loop seal in pressurized water reactors can occur during a small or intermediate break loss-of-coolant accident, causing temporary fuel overheating. Quantification of the accuracy of overheating prediction is of interest in the best-estimate safety analyses, even though the peak cladding temperatures due to the water seal formation in the loop seal seldom approach acceptance criteria as such. The aim of this study was to test and evaluate the accuracy with which the thermal–hydraulic system code nodalizations of the PWR PACTEL predict loop seal clearing in a small break loss-of-coolant-accident test performed with the PWR PACTEL facility. PWR PACTEL is a thermal–hydraulic test facility with two loops and vertical inverted U-tube steam generators. Post-test simulations were performed with the TRACE and APROS system codes. In the post-test simulations, the main events of the transient such as the decrease in the core water level, depressurization of the primary circuit, and the behavior of the water seal formation and clearing in the loop seal were predicted satisfactorily by both codes. However, discrepancies with the experiment results were observed in the analyses with both codes, for example the core temperature excursions were halted too early and the peak temperature predictions were too low. The core water level increase caused by loop seal clearing was overestimated with both codes, and the pressure and temperature were overestimated on the secondary side of the steam generators. Loop Seal 2 was evidently cleared out while Loop Seal 1 remained closed, just like in the experiment. It was noticed that, which of the loop seals clears in the simulations, was sensitive on the nodalization structure and initial conditions. The simulations indicate that the PWR PACTEL nodalizations are capable of simulating loop seal clearing quantitatively well. A quantitative error in the non-conservative direction was observed in peak cladding temperatures in both codes.