Deterministic and probabilistic analysis for the safety of nuclear power plants

Nuclear power plant system codes that model temperature and fluid flow fields are fundamental to the safety assessment of nuclear power plants (NPPs). The primary function of system codes is to predict the physical response of the plant to postulated initiating events (PIEs) due to equipment failures or external hazards. Nuclear safety analysts can understand how an NPP’s structures, systems, and components (SSCs) will behave under normal and abnormal conditions by simulating key parameters like pressure, temperature, and fluid dynamics. The goal is to demonstrate that the safety functions comply with established acceptance criteria and that SSCs act as physical and non-physical barriers protecting the environment, workers, and the public from the harmful effects of ionizing radiation. This doctoral thesis explores the methods and tools used in the nuclear field to determine the safety of both fission and fusion power plants. The main contributions aim to enhance safety by identifying potential weaknesses in the design and thus allowing for improvements in systems and accident prevention measures. The tool used for safety simulations is MELCOR, a 0-D system code, which is capable of simulating thermal-hydraulic parameters during transient sequences and tracking radioactive elements across the various environments within a nuclear facility. Developing system codes for nuclear reactors has primarily focused on Light Water Reactor (LWR) technology, which means these codes are mainly equipped with fission-specific models. However, accident sequences in fusion reactors differ significantly from those in fission reactors, involving physics that is not present in fission-related accident scenarios. Idaho National Laboratories (INL) has implemented a modified version of MELCOR 1.8.6 that incorporates physics relevant to a fusion power plant environment (e.g., cryogenic fluids, tritium transport). Nevertheless, there are still fusion-specific modeling needs. For this reason, during the doctoral activities, efforts have been made to couple MELCOR simulations with external scripts that model phenomena that could not be simulated in the current version of the code. Following the paradigm to broaden the capabilities of standard MELCOR simulation, a new interface has been made between MELCOR and RAVEN, a logic driver for probabilistic analysis. The concept behind this is to explore accident sequences within a probabilistic environment. The final achievement is to provide a Dynamic Probabilistic Risk Assessment (DPSA) tool that is easy to implement with MELCOR and can provide valuable results for severe accident management guidelines. The first main goal of the thesis is to support current design improvements in the ITER Water Cooled Lithium Lead (WCLL) Test Blanket System (TBS). Design basis accidents such as ex-vessel Loss of Coolant Accident (LOCA) or Loss of Flow Accident (LOFA) are studied to evaluate if failure of components could cause an unmitigated release of the radioactive inventory, mainly tritium and Activated Corrosion Products (ACPs). Sensitivity studies are performed to understand different adjustments of the plant systems and their impact on accident consequences. For example, in the case of a LOFA, the current accident mitigation measures can be improved with a different approach to mitigate the overpressure inside the Test Blanket Module (TBM). Safety analyses were performed for the EU-DEMO reactor with a WCLL Breeding Blanket (BB). The current design of the tokamak environments does not include a mitigation system for ex-vessel LOCA. The ex-vessel LOCA for the EU-DEMO reactor studied in this thesis shows the consequence of such a scenario and suggests possible configurations between the tokamak environments. Mitigation systems such as spray and suppression pools are also studied. To enhance the capability of the current MELCOR 1.8.6 fusion version, activity has focused on developing an external model to evaluate the oxidation of tungsten dust with steam. The main result of this study is that most of the hydrogen produced in the case of in-vessel LOCA might come from the oxidation of tungsten dust. Furthermore, a hydrogen explosion risk assessment was conducted, and different configurations of the Vacuum Vessel Pressure Suppression System (VVPSS) were compared. The final result of the thesis is to demonstrate the capabilities of a D-DET interface between MELCOR and RAVEN, a probabilistic driver. DPSA studies have faced a niche application due to their complexity and the scarce availability of platforms to perform such analyses. In terms of applicability, the D-DET interface was developed during the PhD. Defines a straightforward framework for the modeler. Results are shown in terms of Limit Surface to understand the impact of the operator’s action in mitigating the accident sequence. Another result obtained from studying a Station Black Out (SBO) sequence of a Boiling Water Reactor (BWR), is a statistical estimate of the Time to Clad Failure (TCCF). The statistical distribution of TCCF can support the development of Severe Accident Management Guidelines (SAMG) and highlight with confidence the amount of time needed to bring the plant back to a safe state. A final achievement has been the development of a new procedure to estimate the core damage probability. A current method based on D-DET is characterized by an inherent conservatism associated with the discretization of event sampling. The new method tries to solve this issue using a different approach based on a classification function model by a Machine Learning algorithm. In conclusion, this doctoral thesis focuses on the methods and tools adopted in the nuclear sector to analyze the safety of nuclear and fusion power plants. Standard techniques, such as the study of design-based accidents or design-extended conditions, are employed to assess the effectiveness of engineered systems in holding the radiological source term. The tools adopted are labeled conservative system codes able to replicate the thermal-hydraulic parameters during a transient sequence and track the radioactive elements in the different safety barriers of a nuclear facility. The insights of accident sequences are also evaluated using standard and advanced methodologies that account for the probability of occurrence. An important topic in risk assessment is those methods capable of searching the global input of uncertainty and the complete accident scenario to avoid specific gaps that could arise when modeling an accident scenario.