Since the Fukushima Daiichi nuclear accident, there have been ongoing efforts to enhance the modelling capabilities for severe accidents in nuclear power plants. The primary severe accident analysis code used in Canada for its CANDU reactors is MAAP-CANDU (adapted from MAAP-LWR). In order to meet the new requirements that have evolved since Fukushima, upgrades to MAAP-CANDU have been made most recently by the Canadian nuclear industry. While the newest version (i.e. MAAP5-CANDU) offers several important improvements primarily in core nodalization and core collapse modelling, it still lacks mechanistic models for many key thermo-mechanical deformation phenomena that may significantly impact accident progression and event timings. It is also a general consensus that having alternative analysis tools is beneficial in improving our confidence in the simulation results, especially given the complex nature of severe accident phenomena in CANDU and the limited experimental support. This thesis seeks a novel approach to CANDU severe accident modelling by combining the best-estimate thermal-hydraulic code RELAP5, the severe accident models in SCDAP, and several CANDU-specific mechanistic deformation models developed by the author. This work mainly consists of two parts. The first part is focused on the assessment of natural circulation heat sinks following crash-cooldown in the early-phase of a Station Blackout (SBO) accident where fuel channel deformation can be precluded. The effectiveness of steam generator heat removal after crash-cooldown and that of the several water make-up options were demonstrated through the simulation of several SBO scenarios with/without crash-cooldown, sensitivity studies, as well as benchmarking against station and experimental measurements. In the second part, several mechanistic severe accident models were developed to enhance the simulation fidelity beyond the initial steam generator heat sink phase to the moderator boil-off and core disassembly phases. This includes models for predicting the pressure tube ballooning and sagging phenomena during the fuel channel heat-up phase and models for the sagging and disassembly of fuel channel assemblies during the core disassembly phase. After benchmarking against relevant channel deformation experiments, the models were successfully integrated into the RELAP/SCDAPSIM/MOD3.6 code as part of the SCDAP subroutines. The advantage of utilizing a code such as SCDAP is that generic models for fission product release and hydrogen generations, which are well benchmarked, can be directly applied to CANDU simulations. With the modified MOD3.6 code the early-phase SBO simulations were extended to include the later stages of SBO until the calandria vessel dryout. The current modelling approach replaced the simple threshold-type models commonly seen in the integrated severe accident codes such as MAAP-CANDU with more mechanistic models thereby providing a more robust treatment of the core degradation process during severe accident in CANDU. / Thesis / Doctor of Philosophy (PhD)

The safety performance of nuclear power reactors is a very important factor in evaluating nuclear energy sustainability. Improving the safety performance of nuclear reactors can enhance nuclear energy sustainability as it will improve the environmental indicator used to evaluate the overall sustainability of nuclear energy. Great interest is given now to advanced nuclear reactors especially those using passive safety components. Investigation of the improvement in nuclear safety using advanced reactors was done by comparing the safety performance of a conventional reactor which uses active safety systems, such as Pressurized Water Reactor (PWR), with an advanced reactor which uses passive safety systems, such as AP1000, during a design basis accident, such as Loss of Coolant Accident (LOCA), using the PCTran as a simulation code. To assess the safety performance of PWR and AP1000, the “Global Safety Index” GSI model was developed by introducing three indicators: probability of accident occurrence, performance of safety system in case of an accident occurrence, and the consequences of the accident. Only the second indicator was considered in this work. A more detailed model for studying the performance of passive safety systems in AP1000 was developed. That was done using SCDAPSIM/RELAP5 code as it is capable of modelling design basis accidents (DBAs) in advanced nuclear reactors.