izpis_h1_title_alt

Use of hybrid methods for neutron transport and sensitivity analysis
ID Kos, Bor (Author), ID Kodeli, Ivan Aleksander (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (22,24 MB)
MD5: 4CC03533FBABF8E826F3745E4CB81ECD

Abstract
Accurate and efficient determination of uncertainties of physical quantities is a necessary step of any experiment or computational analysis. Uncertainties in particle transport simulations can be divided by type, statistical or systematic, and by origin i.e. uncertainties in nuclear data and uncertainties originating from approximations. Specifically in neutron transport simulations of fusion and fission shielding applications the approximations can be attributed to geometry and material specification, numerical (discretization of the phase space) approximations of deterministic transport methods and statistical uncertainty of stochastic transport methods. Accurate knowledge of all uncertainties related to transport simulations leads to reduced safety margins and consequently to more efficient experimental and commercial designs. The thesis is focused on developing new methods for efficient propagation of nuclear data uncertainties and reducing the statistical uncertainties of stochastic transport methods with hybrid transport methodologies. The first chapter of the thesis gives an overview of all theoretical aspects of neutron transport, variance reduction, hybrid methodologies and nuclear data (ND) sensitivity and uncertainty (SU) quantifications. The second chapter contains the description and evaluation of computer codes used in the thesis. An extensive evaluation and validation of the ADVANTG hybrid code is presented in this part. Besides the deterministic approach to ND uncertainty quantification the stochastic approach is also evaluated and used for validation purposes. In the last chapter the main contributions to the scientific field by the thesis author are presented. A large emphasis is given to the work that was done to develop a user friendly and efficient code system ASUSD (ADVANTG + SUSD3D) for ND S/U quantification. The code system enables ND S/U analysis of complex shielding experiments which was previously cumbersome, computationally expensive or even impossible. The ASUSD code system was validated and applied to two complex fusion experiments including the challenging nexp streaming benchmark experiment. Moreover ASUSD was used to asses the effect of self-shielded cross sections on the effectiveness of variance reduction parameters. The final contribution to the scientific field of hybrid particle transport is the development of a new method for determining variance reduction parameters based on sensitivity profiles - saber. SABER preforms similarly or better than the current established methodologies.

Language:English
Keywords:hybrid transport, variance reduction, nuclear data, sensitivity, uncertainty, shielding, fusion, fission, total Monte Carlo (TMC), MCNP, ADVANTG, Denovo, SUSD3D, SANDY
Work type:Doctoral dissertation
Typology:2.08 - Doctoral Dissertation
Organization:FMF - Faculty of Mathematics and Physics
Year:2020
PID:20.500.12556/RUL-114388 This link opens in a new window
COBISS.SI-ID:3411300 This link opens in a new window
Publication date in RUL:26.02.2020
Views:1874
Downloads:522
Metadata:XML DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:Slovenian
Title:Uporaba hibridnih metod za transport nevtronov in analizo občutljivosti
Abstract:
Točno in učinkovito določanje negotovosti fizikalnih količin je nujen korak vsakega eksperimenta ali računalniške simulacije. Negotovosti na področju simulacij transporta delcev lahko delimo glede na tip, na sistematične oziroma statistične, in glede na izvor npr. negotovosti v jedrskih podatkih oziroma negotovosti, ki izvirajo iz približkov. Specifično pri simulacijah transporta nevtronov fuzijskih in fisijskih aplikacij radiološkega ščitenja lahko približke dodelimo specifikacijam geometrijskega modela in materialne sestave, numeričnim aproksimacijam (diskretizacija faznega prostora) pri determinističnih metodah in statističnim negotovostim, ki se pojavijo pri stohastičnih metodah za transport delcev. Točno poznavanje vseh negotovosti povezanih s simulacijami transporta delcev privede do zmanjšanja varnostnih faktorjev in posledično bolj učinkovitih eksperimentalnih in komercialnih zasnov. Disertacija je osredotočena na razvoj novih metod za učinkovito propagacijo negotovosti jedrskih podatkov in zmanjšanje statistične negotovosti pri stohastičnih simulacijah transporta delcev s hibridnimi metodologijami. Prvo poglavje disertacije povzema vse teoretične vidike transporta nevtronov, redukcije variance, hibridnih metodologij ter določanje občultjivosti in negovosti (O/N) zaradi jedrskih podaktov (JP). Drugo poglavje vsebuje opis in evalvacijo programov, ki so bili uporabljeni v disertaciji. Predstavljena je tudi obsežna evalvacija in validacija hibridne kode ADVANTG. Poleg determinističnega pristopa k O/N zaradi JP, je za namene validacije predstavljen tudi stohastični pristop k določevanju negotovosti zaradi JP. V zadnjem poglavju so predstavljeni glavni doprinosi avtorja disertacije znanstvenemu področju. Velik poudarek je namenjen predstavitvi razvoja uporabniku prijaznega in učinkovitega programskega paketa ASUSD (ADVANTG + SUSD3D) za O/N zaradi JP. Programski paket omogoča analizo O/N zaradi JP kompleksnih aplikacij radiološkega ščitenja, ki so bile do sedaj okorne, računalniško potratne ali celo nemogoče. Paket ASUSD je bil validiran na dveh referenčnih eksperimentih in apliciran na dva kompleksna fuzijska problema, vključno z zahtevnim referenčnim eksperimentom strujanja, JET3-NEXP. Poleg tega, je bil paket ASUSD uporabljen za preučevanje efekta samoščitenih JP na učinkovitost parametrov za redukcijo variance. Zaključni doprinos znanstvenemu področju hibirdnega transporta delcev je razvoj nove metode za določevanje parametrov za redukcijo variance na podlagi občutljivostnih profilov - SABER. SABER deluje enako dobro oziroma bolje kot trenutno uveljavljena metodologija.

Keywords:hibridni transport, redukcija variance, jedrski podatki, občuljivost, negotovost, ščitenje, fuzija, fisija, totalni Monte Carlo, MCNP, ADVANTG, Denovo, SUSD3D, SANDY

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back