VERJETNOSTNE VARNOSTNE ANALIZE JEDRSKE ELEKTRARNE V ZAUSTAVITVI

ANTONČIČ, MITJA (avtor), Čepin, Marko (mentor)

Pričujoča magistrska naloga obravnava verjetnostne varnostne analize jedrske elektrarne v zaustavitvenih stanjih. Verjetnostne varnostne analize so namenjene ocenjevanju in izboljšanju varnosti kompleksnih sistemov, tudi jedrskih elektrarn. Skozi analizo izvemo možne neželene dogodke, do katerih lahko pride v sistemu, verjetnost nastopa teh dogodkov, način njihovega razvoja in končne posledice. Na podlagi analize lahko določimo pomembnost posameznih komponent, kar je koristna informacija pri načrtovanju vzdrževalnih del ter načrtovanju nadgradenj v sistemu. V začetku naloge je predstavljenih nekaj osnov verjetnostne teorije, splošen opis verjetnostnih varnostnih analiz in opis izvedbe analiz. Uporabljena je metoda dreves odpovedi in dreves dogodkov. S prvimi so modelirani posamezni sistemi elektrarne, drugi pa ponazarjajo razvoj neželenega začetnega dogodka. Nato je opisana izvedba verjetnostnih varnostnih analiz jedrske elektrarne v obratovanju pri nizki moči in zaustavitvi, čemur sledi opis razvitega modela. Model je bil razvit znotraj programskega orodja RiskSpectrum PSA. Model temelji na podatkih ameriške elektrarne Surry ter na obstoječem modelu, namenjenem varnostnim analizam elektrarne pri delovanju na polni moči. Obravnava 15 različnih obratovalnih stanj elektrarne: od zniževanja moči, prek vroče pripravljenosti, vroče zaustavitve, hladne zaustavitve, menjave goriva in v obratnem vrstnem redu do obratovanja na polni moči. Neželeno končno stanje je poškodba sredice reaktorja. Sledi predstavitev rezultatov, ki prikazujejo nivo tveganja poškodbe sredice v posameznem obratovalnem stanju elektrarne. Rezultati prikazujejo časovno uteženo tveganje, prispevek posameznega stanja k skupnemu tveganju ter pogojno tveganje v primeru, da bi vsa stanja trajala enako dolgo. V zaključku sledi komentar rezultatov in komentar modela ter njegovih pomanjkljivosti. Znotraj modela so zajeti le notranji dogodki v elektrarni, zunanjih (npr.: potres, padec letala) pa izdelani model ne obravnava. Predstavljeni rezultati pričakovano prikazujejo višje tveganje v času obratovanja pri znižanem nivoju hladila in najnižje tveganje v času, ko je reaktorski bazen poplavljen. Model je možno nadalje izboljšati z ustreznim ovrednotenjem posameznih vrednosti (npr. človeški faktor, časovne konstante), kjer te verjetnosti predstavljajo bistven doprinos k skupnemu tveganju. Ker se delo naslanja predvsem na tujo literaturo, se je bilo v besedilu nemogoče povsem izogniti tujim izrazom. Zato imajo posamezni slovenski prevodi in krajšave zaradi lažjega razumevanja v oklepaju podane angleške ustreznice.

• Jezik: Slovenski jezik
• Ključne besede: verjetnostne varnostne analize, jedrska elektrarna, drevo odpovedi, drevo dogodkov, tveganje
• Vrsta gradiva: Magistrsko delo/naloga
• Organizacija: FE - Fakulteta za elektrotehniko
• Leto izida: 2016
• PID: 20.500.12556/RUL-84069
• Datum objave v RUL: 07.07.2016

Sekundarni jezik

• Jezik: Angleški jezik
• Naslov: PROBABILISTIC SAFETY ASSESSMENT OF A NUCLEAR POWER PLANT IN ITS SHUTDOWN

Izvleček

The following master thesis presents Probabilistic Safety Assessment (PSA) of nuclear power plants during low power and shutdown operating states. Purpose of Probabilistic Safety Assessment is assessment and improvement of safety of complex systems, such as nuclear power plant. Analysis reveals possible undesired events that could occur in the system, likelihood of their occurrence, the way of their development and their consequences. Analysis also reveals importance of individual components of the system, which serves as a base for planning maintenance and upgrades in the system. In the beginning, the basics of probability theory are presented together with general description of Probabilistic Safety Assessment and description of implementation of PSA. Event tree and fault tree approach is used for model development. Connections between different components and systems are modelled with fault trees, while in event trees development of undesired initial events is presented. This is followed by description of Low Power and Shutdown PSA of nuclear power plants. It is supported with description of a computer model for performing such analyses. Model was developed with RiskSpectrum PSA, a powerful Probabilistic Safety Assessment computer program. Model is based on data from the US nuclear power plant Surry and existing model for nuclear power plant safety analysis during full power operation. Model consists of 15 sub models for 15 different plant operating states from lowering power through hot standby, hot shutdown, cold shutdown, refuelling and then back towards full power again. Core damage frequency is the main risk measure. Later on some results are presented. Core damage frequencies (CDF) are presented in both numerical and graphical manner. Firstly time weighted CDF are shown, than contribution of different plant operating states (POS) to overall CDF and in the end conditional CDF of each POS is graphically presented. In the end, there are some comments of results and model presented together with its limitations and possible future upgrades. Developed model covers only internal initiating events of nuclear power plants without external ones (e.g. earthquakes, airplane crash). Results show higher risk during mid-loop states and lowest risk during refuelling, when reactor cavity is flooded. Model could be further enhanced by additional analysis of the most important risk contributors (Human factor, time constants).

• Ključne besede: Probabilistic Safety Assessment, nuclear power, fault tree, event tree, risk