Well-known accident at the Fukushima NPP, caused by devastating earthquake, was followed, in the whole world, by evaluations and analyses, either from nuclear industry or/and national nuclear regulators point of view. Among other international organisations, ENSREG (European Nuclear Safety Regulators Group) initiated so called “stress test”, asking the EU countries to re-evaluate nuclear safety of existing NPPs towards rare and extreme external events such as seismic events of intensity above plant’s design bases. Such evaluations resulted with numerous changes and revisions related to nuclear power plants design requirements (e.g. IAEA SSR-2/1 “Safety of Nuclear Power Plants: Design”, WENRA RHWG Report “Safety of new NPP designs”, EUR Revision E, etc.) for either existing nuclear power plants or new ones. One of the biggest news in nuclear power plant design requirement terminology is the concept of so called „Practical Elimination“. The accident sequences with a large or early release can be considered to have been practically eliminated:  if it is physically impossible for the accident sequence to occur or  if the accident sequence can be considered with a high degree of confidence to be extremely unlikely to arise. In each of the mentioned cases the demonstration should show sufficient knowledge of the accident condition analysed and of the phenomena involved, substantiated by relevant evidence. However, industry and regulators did not clearly prescribe a methodology for demonstrating the practical elimination principle using deterministic or / and probabilistic safety analyses. The main goal of research covered by this dissertation is focused on development of methodology for safety demonstration of practical elimination of significant fuel damage in a spent fuel pool (SFP) as extremely unlikely event with a high degree of confidence. Practical elimination of an accident sequence cannot be claimed solely based on compliance with a general cut-off probabilistic value. Even if the probability of an accident sequence is very low, any additional reasonably practicable design features, operational measures or accident management procedures to lower the risk further should be implemented as it was demonstrated in the provided thesis. The thesis consists of seven chapters. Chapter 1 serves as an introduction and general overview of the problem as described above. Chapter 2 describes general methodology for qualitative and quantitative assessment of the risk of damage to nuclear fuel in SFP and the assessment of the possible conditions of the power plant in the case of a complete loss of coolant and physical damage to the fuel cladding in the SFP. It provides categorization of the initial events of all possible internal and external scenarios for an individual structure (considering two main locations with nuclear fuel, reactor and SFP) as well as combined scenarios (influence of hazards and scenario propagation from one structure to another). Chapter 3 provides more detailed elaboration of methodology including the following items:  definition of the operating conditions of the spent fuel pool cooling system;  definition of potential initiating events (IE – initiating events, e.g., loss of cooling system, loss of water inventory or loss of power supply) that can lead to loss of coolant or damage to fuel rod claddings of spent fuel;  estimation of the time windows for each initiating event in which there would be boiling of the coolant in the spent fuel pool and complete loss of the coolant;  examples of created event trees (ET) and fault trees (FT) in the probabilistic safety analysis (PSA) model for defined initiating events, based on the analyses of operable systems and their safety functions including determination of applicable SFP PSA model parameters;  examples of reliability analysis of operator actions in the case of an accident (Human Reliability Analysis);  qualitative and quantitative (probabilistic) risk assessment of nuclear fuel damage in the SFP located in the fuel handling building (FHB) due to internal events and/or fuel damage in the reactor core in the containment; The probabilistic quantification was done for two general main events: SFP boiling frequency (SFPB) and spent fuel uncovery (SFU). For the purpose of risk analysis and quantification, the SFP conditions are represented by a set of SFP systems / components / structures configurations and thermal loads. The quantification analysis of the frequency of SFU on individual initiators shows that the biggest contribution comes from the loss of SFP inventory, large and small leakages. The importance of loss of coolant events (compared to loss of cooling events) should come as no surprise, given that the loss of cooling events are characterized by long time windows (which are governed just by boiling and evaporative losses), which can be significantly reduced by loss of coolant (where leakages exceed loss of water by evaporation). Results of demonstration SFP PSA model are comparable with similar Electric Power Research Institute (EPRI) study results (between 1E-7/year and 1E-10/year for different PWR plants) Chapter 4, taking into account the PSA results provided in Chapter 3, discusses the performed analysis of the effectiveness of SFP spray cooling during SFP loss of coolant accidents (as the last barrier in the case of assumed loss of power, fire, loss of cooling, loss of coolant, seismic event or airplane crash). The MELCOR 2.1 model of NPP Krsko SFP was developed and tested for cases of loss of cooling accidents. The simple spray system with spray nozzles distributed at specified location at the top of the pool was added to the model. According to analysis results, spray nozzles that are installed are able to limit or delay long-term heat-up of the spent fuel, but in the case of late actuation it is possible to have temporary high oxidation rates and corresponding production of hydrogen, requiring additional operator actions such as actuation of emergency ventilation in the FHB. The performed calculations are conservative and use the ring-based approach to SFP modelling. Analysis shows that efficiency of spray system is eventually capable of limiting fuel temperature increase in SFP, but depending on time of actuation and droplets size can cause certain hydrogen generation. Main focus of the dissertation is on the assessment of potential impact of combustible gases from reactor core damage on risk from damage to spent fuel pool located outside containment, as discussed in Chapter 5. An important element of the SFP risk analysis is the assessment of a possible impact of a severe accident in the reactor, which is in the containment, to SFP outside the containment. Chapter 5 analyses deal with one particular aspect of this problem and evaluate possible direct or indirect impact of the combustible gases generated in the course of reactor accident and released out of the containment. The methodology for assessment of possible hydrogen presence in the containment annulus, its flammability, and leakages through the penetrations toward the FHB in the case of a long-term station blackout (SBO) without successful restoration of the core cooling is described. Generally, the dependency between the reactor response and the SFP response to postulated initiators may primarily arise due to the following: 1. Simultaneous failures related to initiating event (e.g., loss of offsite power, seismic events, presence of Design Extension Conditions (DECs), etc.); 2. Reactor severe accident conditions which propagate to adverse conditions affecting the FHB/SFP structure or the SFP cooling / make-up equipment (“make-up” equipment here generally referring to the plant systems provided for adding water into the SFP in order to preserve adequate water inventory in SFP in the case of its loss due to ruptures (loss of inventory events) or evaporation (total loss of cooling events)). The major adverse conditions from the PSA Level 1 and Level 2 studies that affect the SFP event tree structure include the following:  Hydrogen release that could result in deflagration events and structures or electrical / mechanical equipment failures;  Containment failures that cause similar effects;  Fission product releases that inhibit or preclude access to the areas needed for local alignments;  Failure of all the installed equipment may force the Technical Support Centre (TSC) staff to decide how to prioritize the use of any remaining portable equipment. Decision for use in one application (e.g. reactor accident mitigation) may preclude its subsequent realignment to SFP due to local environmental conditions. Thesis deals with one particular aspect of this problem and evaluates the possible propagation of the combustible gases generated in the course of reactor accidents out of the containment. The focus is on the evaluation of potential combustible gas leaking into the annulus and FHB (where, unlike the containment, it is not recombined as PARs (Passive Autocatalytic Recombiners) are not installed there) and its potential subsequent deflagration and impact on SFP. The leakage path can be from the containment to the annulus (and then to surrounding buildings or the environment), directly to the environment or directly to one of the neighbouring buildings. The Krško NPP’s MAAP5.03 state-of-the-art model was used for analyses in the scope of the work presented (and original Krško NPP PSA Level 2 Studies), including all relevant plant-specific engineered safety systems such as emergency core cooling, power-operated relief valves, containment sprays, and fan coolers. Taking into account the results of the previous Krško PSA Level 2 studies, the basic scenario of SBO (Station Black Out) sequence was chosen for the assessment of conditions in the annulus. Analysis takes into account 5 different sensitivity analyses related to hydrogen production and distribution, including flammability conditions in the containment annulus. Sensitivity analyses take into account various changes in initial and boundary conditions relating to the operability of Passive Containment Filtered Vent (PCFV) systems and assumed leakages from the containment to the annulus, from the containment to the atmosphere and from the annulus to the atmosphere. It is pointed out that sensitivity analyses did not assume any operator actions. When interpreting the results, calculation uncertainties need to be taken into account: for example, no results were available from any full-scale test of combustible gases leakage into the annulus and the assessment of the deflagration conditions. Also, it needs to be considered that conservative assumptions were taken, e.g. large containment breach (0.1 m2) assumed at the containment design pressure (0.41 MPa) rather than at the ultimate stress pressure (0.735 MPa). Nevertheless, the results of analyses from postulated scenarios show that in the case of a DEC event (e.g. assuming a seismic initiator above the design value), when leakages can be assumed conservatively much higher than design leakages (or when leakages are increasing before reaching the ultimate stress pressure), there is very small possibility for combustible gasses deflagration and it is reasonable to recommend to assure containment annulus venting as soon as possible. Although the MAAP5.03 analyses demonstrate that deflagration risks for assumed scenarios are very low (combustible gasses volume fraction for all cases was bellow deflagration limit (<4%) before containment failure or PCFV actuation), it is reasonable to consider potential energetic containment or annulus failure, with impact on FHB (and SFP PSA Event Tree logic). It may be reasonably considered that the probability of the impact on FHB would decrease by an impact area factor of 0.07 (ratio of the relevant wall surface over the total wall surface) for evaluation of consequences of such event in FHB and regarding the systems, structures and components located there, for the protection of safety functions related to the SFP integrity (maintaining the SFP inventory and decay heat removal). Logical and probabilistic model for quantification of impact of reactor core damage sequences on SFP was developed and discussed in details. In a detailed PSA model, an SFP event tree would be developed specifically for each of defined end-states form major logical model, reflecting specific damages to the SFP structures and / or mitigation systems imposed by the considered end-state. Thesis did not attempt to characterize the probabilities of particular sequences. The presented assessment in Chapter 5 demonstrates how important it can be to adequately address the candidates for a high-level mitigative action (CHLA) related to ventilation of the auxiliary buildings in the plant-specific SAMGs, because it is impossible to eliminate the potential for a breach in the containment at the onset of the accident (or postulated DEC) or as a consequence of harsh conditions that develop inside the containment. This is also important because the presence of flammable volume fractions in the annulus or FHB is not monitored from the main control room. Such conditions show that during periods with relatively high combustible gases in the annulus combustion is prevented by an inert environment (due to steam (and CO2 from molten core – concrete interaction (MCCI)) volume fraction above 55 %) and low volume fraction of oxygen (less than 20 % is needed for deflagration or detonation). Particularly, the Chapter 5 recommends that the best estimate sensitivity deterministic analyses (by MAAP or MELCOR) would be needed to evaluate more realistically containment leakage distribution, including a more detailed presentation of connections between FHB/SFP and the containment, as well as a more detailed model of adjacent buildings to decrease all postulated conservatisms Chapter 6 discusses the bases for creating an application for severe accident management tools to help plant staff (technical support centre and main control room) on various available technical measures (or the use of individual operator actions to mitigate consequence of severe accident). Chapter 6 summarizes the previous work on EU NARSIS project and development of SEVERA tool which was focused just on implementation of Severe Accident Management Guidelines (e.g. depressurizing reactor coolant system (RCS), inject water in the RCS and controlled conditions in the containment) during severe accident progression events in the containment. SEVERA tool was developed based on DEX multi-variants decision making tool (developed by Institute Josef Stefan in Ljubljana). Usage of SEVERA was successfully demonstrated. However, work performed during preparation of thesis defines the major improvements in the logic model of SEVERA and additional attributes needed in hierarchical model to be able for decision making process to relate the interaction of core damage states in the containment with hazards related to fuel in SFP. Conclusions in Chapter 7 summarize the compliance of dissertation with major goals and underline the scientific contributions resulting from this research: 1. Methodology for systematic qualitative and quantitative risk assessment of hazards impact on SFP was developed and usability demonstrated; 2. Probabilistic (very low frequency) and deterministic criteria (no fuel uncovering in the first 72 hours) for demonstration of physical elimination of event consequences were defined and successfully compared with results of chosen probabilistic and deterministic analyses; 3. All the necessary bases and attributes have been created for upgrading the decision-making process during the prioritization of actions in the case of unlikely events that would threaten both the primary circuit of the power plant and the pool for spent nuclear fuel (like a SEVERA application developed under the EU NARSIS project).