AN ASSESSMENT OF HEAT TRANSFER RESULTS FOR REFLUX CONDENSATION IN APPLICATION TO CANDU SEVERE ACCIDENT CONDITIONS

Abstract

Severe accident progression in a nuclear reactor (including CANDU: CANada Deuterium Uranium, pressurized heavy water reactor) is complex and involves many phenomena (fuel heat-up and meltdown, core disassembly, fission product release, hydrogen production, etc.). While significant progress has been made in understanding these phenomena, some processes are not yet fully understood. A major area of concern is the possibility of nuclear fuel heat-up due to the decay heat, which is not removed due to insufficient cooling, and failure of fuel channels and the resulting core disassembly. Reflux condensation (RC) is a complex phenomenon related to heat-up of the reactor primary coolant side while a secondary side coolant heat sink is available. RC can impact the progression of the fuel (and fuel channel) heat up and accident progression event timings (fuel channel dry-out, onset of severe damage, and failure).

The RC process is becoming more important due to the recent CANDU plant safety system upgrades (such as external water make-up to the steam generator (SG) secondary side), as a result of the Fukushima 2011 nuclear accident, which better ensure that the SG heat sink is maintained and supports conditions for RC.

The MAAP-CANDU code is the primary severe accident analysis tool for CANDU plants. A study has been carried out at CNL to gain more confidence in the robustness of the MAAP-CANDU code models, in predicting the heat and mass transfer for RC phenomena. Comparisons were made between the experimental data from RC tests conducted at McMaster University and Computational Fluid Dynamics (CFD) simulations to benchmark the MAAP-CANDU RC models. Additionally, direct benchmarking of the experiments to MAAP-CANDU were conducted.

Experimental studies directed towards improved understanding of the distribution of coolant flow in reactor headers and feeder pipes resulting from the RC in SG tubes are being carried out at the CNL’s Header Test Facility (HTF).

Some results of these experimental studies and benchmarking that could suggest upgrades of the related MAAP-CANDU models are explored in this paper. Further steps/efforts for experiments and models benchmarking are also outlined.