Search

During a severe accident in a light water reactor nuclear power plant, hydrogen may be generated, and its combustion could threaten the integrity of the nuclear power plant containment, which could lead to release of radioactive material into the environment. The study of hydrogen distribution in the containment is important to predict the occurrence of regions with flammable mixture, in order to effectively install hydrogen mitigation systems. A theoretical model to describe the homogenization of the layered atmosphere due to the vertical injection in an enclosure was developed. In linear two-equation eddy viscosity turbulence models, constant values of turbulent Schmidt and Prandtl numbers are usually used. It was shown that constant values fail to reproduce some phenomena observed in experiments. For the purposes of this work, firstly, a local Froude number is introduced, which can be directly used in computational fluid dynamics calculations. Secondly, a model that prescribes turbulent Schmidt and Prandtl numbers dynamically, which uses local Froude number and vertical velocity, is proposed. Non-essential parts of the experimental facilities, which do not contribute much to the mixing process, were identified according to experimental results. Accordingly, axisymmetric two-dimensional and three-dimensional numerical models of cylindrical vessels were developed, and the effects of numerical domain reduction were studied. Since OpenFOAM is a multipurpose computational fluid dynamics code, it was first adapted to simulate the considered cases. An additional term accounting for enthalpy diffusion due to compositional changes was implemented into the energy equation, and a term describing the molecular diffusion had to be added to the convection-diffusion equation for mean mass fraction. The developed physical model for simulation of mixing in nonhomogeneous atmosphere in large enclosures was verified and validated on fourteen different experiments performed in three experimental facilities. The comparison with experimental results showed that the selected modelling approach is able to correctly predict the helium distribution and temperature values. Furthermore, with the proposed model the successful modelling of breaking-up of atmosphere stratification, induced by vertical injection, was extended to isothermal conditions with high mass inflow. Med težko nesrečo v lahkovodnem jedrskem reaktorju lahko pride do nastanka vodika, njegovo zgorevanje pa lahko ogrozi celovitost zadrževalnega hrama jedrske elektrarne, kar bi lahko vodilo v izpust radioaktivnih snovi v okolje. Študija porazdelitve vodika v zadrževalnem hramu je pomembna za napovedovanje območij z vnetljivo mešanico. Razvit je bil teoretični model za opis homogenizacije razslojene atmosfere zaradi navpičnega vbrizgavanja v eksperimentalni napravi zadrževalnega hrama. V turbulentnih modelih, ki temeljijo na modeliranju vrtinčne viskoznosti se običajno uporabljajo konstantne vrednosti turbulentnega Schmidtovega in Prandtlovega števila. Pokazano je, da konstantne vrednosti teh števil ne morejo pravilno reproducirati nekaterih fizikalnih pojavov opaženih v eksperimentih. Za namene tega dela je najprej definirano lokalno Froudovo število, ki se lahko neposredno uporablja v programih za računsko dinamiko tekočin. Poleg tega je predlagan model za dinamično predpisovanje turbulentnega Schmidtovega in Prandtlovega števila, ki uporablja lokalno Froudovo število in navpično hitrost. Nepomembni deli eksperimentalnih naprav, ki ne prispevajo veliko k procesu mešanja, so bili identificirani s pomočjo eksperimentalnih rezultatov in rezultatov simulacij. V skladu z ugotovitvami so bili razviti poenostavljeni osno simetrični dvodimenzionalni in tridimenzionalni numerični modeli cilindričnih posod ter proučeni učinki redukcije numerične domene. Ker je OpenFOAM večnamenski program za računsko dinamiko tekočin, je bil najprej prilagojen za modeliranje obravnavanih primerov. V energijsko enačbo je bilo potrebno implementirati dodatni izraz, ki upošteva difuzijo entalpije zaradi sprememb sestave, konvekcijsko-difuzijski enačbi za masni delež plina pa je bilo potrebno dodati člen, ki popisuje molekularno difuzijo. Razviti fizikalni model za simulacijo mešanja v nehomogeni atmosferi je bil preverjen na štirinajstih eksperimentih izvedenih v treh različnih eksperimentalnih napravah. Primerjava z eksperimentalnimi rezultati je pokazala, da razviti fizikalni model pravilno napove porazdelitev helija in temperaturo znotraj zaprte posode. S predlaganim modelom za dinamično predpisovanje turbulentnih števil je bilo uspešno modeliranje erozije razslojene atmosfere z navpičnim vpihovanjem razširjeno na izotermne pogoje z visokimi hitrostmi vpihovanja, pri katerih pride do nastanka Kelvin-Helmholtzeve nestabilnosti.

During a hypothetical severe accident in a light water nuclear power plant, the molten reactor core may come in contact with the coolant water. One of the consequences can be a vapour explosion. Typically in nuclear safety, the vapour explosions are mostly analysed in the melt jet-coolant pool geometry. In the stratified melt-coolant configuration, which was less analysed, a layer of melt lies below a layer of coolant. The stratified melt-coolant configuration was believed to be incapable of producing a strong energetic interaction between the melt and coolant.. This belief was based on the conclusions from the past theoretical and some experimental research, where the lack of the interfacial instabilities and consequently the lack of an explosive premixture was determined. In the recent fuel-coolant interaction experiments performed in the stratified configuration at the SES and PULiMS facilities (KTH, Sweden), a premixed layer formation of ejected melt drops in water was clearly visible and was followed by strong spontaneous vapour explosions. The purpose of our research was to improve the knowledge, understanding and modelling of the fuel-coolant interaction phenomena in the stratified configuration. Firstly, the phenomenon of the vapour explosion is presented. An overview of the experimental work is given. The SES and PULiMS experiments (KTH, Sweden) and other experiments of vapour explosion and premixed layer formation in stratified configuration as well as some other relevant experiments are presented. Existing modelling approaches for mixing in case of the vapour explosions in the stratified configuration are reviewed. Possible mechanisms for the premixed layer formation are discussed and assessed according to their relevance to the reviewed experiments. Based on the visual observations and some available mechanisms from the literature, models for the premixed layer formation are presented. Due to the uncertainties and the lack of detailed information, two different approaches are followed in the modelling. Both approaches are based on the bubble collapse mechanism, where coolant micro-jet is formed. The first modelling approach describes the melt drops ejection due to the coolant micro-jet penetration in the melt and its vaporization. The second modelling approach describes the melt drops ejection due to the pressure perturbation, when the coolant micro-jet hits the melt surface. Both models and experimental observations are discussed and joined into the third model, which describes the considered phenomena the best. The developed model is implemented into the MC3D code (IRSN, France) and validated against the experimental results. To enable the implementation of the proposed model into the Eulerian fuel-coolant interaction codes, a description of melt drops with a two-melt-drop-group approach is introduced. The behaviour of the implemented model is verified with numerical and physical tests. The implementation of the model into the fuel-coolant interaction code enables the simulation of the observed fuel-coolant interaction in the stratified configuration. The presented analyses demonstrate the model capability to describe the premixed layer formation in a qualitative agreement with the available experimental data. The indirect comparison of the simulated premixed layer with the SES S1 and PULiMS E6 experiments via the strength of vapour explosion shows underestimation in simulations’ explosion strength. This underestimation is further discussed and other possible contributions, e.g. contribution from the melt jet break-up and mixing during the explosion are highlighted. Hipotetična težka nesreča v jedrski elektrarni lahko vodi do poškodb sredice, vključno s taljenjem sredice. Če staljena reaktorska sredica pride v stik z vodo lahko pride do parne eksplozije. V reaktorskih varnostnih analizah se parne eksplozije večinoma analizira v geometriji izliva curka taline v bazen hladila. V razslojenih razmerah, ki so bile deležne manj pozornosti, pa leži plast taline pod plastjo hladila. Za razslojene razmere taline in hladila se je predpostavljalo, da ne morejo privesti do močnih interakcij med talino in hladilom. Ta hipoteza je bila osnovana na teoretičnih in eksperimentalnih študijah, kjer niso opazili površinskih nestabilnosti in posledično eksplozivne mešanice taline in hladila. V pred kratkim opravljenih eksperimentih interakcije taline s hladilom v razslojenih razmerah na eksperimentalnih napravah SES in PULiMS (KTH, Švedska) je bil jasno opažen nastanek mešalne plasti kapljic taline v vodi, kateremu je sledila močna spontana parna eksplozija. Namen našega raziskovalnega dela je doprinos k znanju, razumevanju in modeliranju interakcije taline s hladilom v razslojenih razmerah. Najprej je predstavljen pojav parne eksplozije. Opravljen je pregled eksperimentalnega dela. Predstavljena sta pred eksperimenta PULiMS in SES (KTH, Švedska) ter nekateri ostali eksperimenti, povezani s parnimi eksplozijami in mešalno plastjo v razslojenih razmerah. Opravljen je pregled modeliranja mešanja pri parnih eksplozijah v razslojenih razmerah. Obravnavani so možni mehanizmi za nastanek mešalne plasti in ocenjeni glede na njihovo relevantnost za posamezne eksperimente. Na podlagi opazovanj in nekaterih razpoložljivih mehanizmov iz literature so predstavljeni modeli za nastanek mešalne plasti. Zaradi negotovosti in pomanjkanja podrobnih informacij pri modeliranju uporabimo dva različna pristopa. Oba pristopa slonita na mehanizmu kolapsa mehurčkov, kjer nastanejo mikrocurki hladila. Prvi pristop k modeliranju predpostavlja izmet kapljic taline zaradi vdora mikrocurkov hladila v talino ter njihovega uparjanja, drug pristop k modeliranju pa izmet kapljic taline zaradi tlačne motnje ob udarcu mikrocurka hladila ob talino. Oba modela ter rezultati eksperimentov so obravnavani in združeni v tretji model, ki najbolje opisuje obravnavane pojave. Razviti model je vključen v računalniški program MC3D (IRSN, Francija) in validiran na eksperimentalnih rezultatih. Za vključitev predlaganega modela v računalniški program Eulerjeve narave za interakcijo taline s hladilom se vpelje dvogrupni pristop za popis kapljic taline. Pravilno obnašanje modela je verificirano z numeričnimi in fizikalnimi testi. Vključitev modela v računalniški program za interakcijo taline s hladilom nam omogoča simulacijo opažene interakcije taline s hladilom v razslojenih razmerah. Predstavljena analiza pokaže zmožnost modela za popis nastanka mešalne plasti v kvalitativnem ujemanju z razpoložljivimi rezultati eksperimentov. Posredna primerjava simulirane mešalne plasti v eksperimentalnih testih SES S1 in PULiMS E6 preko moči parne eksplozije pa nakaže podcenitev moči eksplozije v simulacijah. Razprava zajema opažene razlike in osvetli nekatere možne dodatne prispevke, kot sta prispevka zaradi razpada curka taline in mešanje med samo eksplozijo.