| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

Show document Help

Title:Analiza vpliva prostorske diskretizacije metode končnih elementov na natančnost numeričnih rezultatov
Authors:ID Šantl, Jure (Author)
ID Ren, Zoran (Mentor) More about this mentor... New window
ID Borovinšek, Matej (Comentor)
Files:.pdf MAG_Santl_Jure_2015.pdf (2,86 MB)
MD5: DAFB58D89ED812F6ECE4B48DDBBA6D4A
 
Language:Slovenian
Work type:Master's thesis/paper
Typology:2.09 - Master's Thesis
Organization:FS - Faculty of Mechanical Engineering
Abstract:Računalniške simulacije izvajamo tako, da model diskretiziramo (razdelimo na elemente). Na ta način iščemo približno rešitev zadanega problema. Kvaliteta končnih elementov bistveno vpliva na natančnost rešitev. Opisal sem, kako geometrijski tip, integracijska shema, geometrijska oblika ter velikost končnih elementov vplivajo na natančnost numeričnih rezultatov. Opisal sem probleme, ki se pojavljajo pri izbiri polne oziroma reducirane integracije. Primernost oblike končnih elementov najpogosteje vrednotimo na podlagi razmerja med stranicami elementa, velikostjo notranjih kotov in Jacobijeve determinante. Raziskal sem metode za ocenjevanje napake diskretizacije. Ena izmed njih je Richardsonova ekstrapolacija, ki sem jo podrobneje raziskal. To je metoda, ki na podlagi treh rešitev na različno gostih mrežah oceni ekstrapolirano vrednost spremenljivke (npr. napetosti). Tako pridobimo oceno vrednosti spremenljivke, ki bi jo dobili na mreži z neskončno malimi elementi (velikost elementa je 0). Na tem temelji indeks konvergence mreže GCI, ki ga uporabimo kot oceno napake diskretizacije. Ta oceni, znotraj katerega intervala leži s 95 % verjetnostjo natančna vrednost matematičnega modela. Izdelal sem dodatek za programski paket Abaqus, ki temelji na tej metodi. Dodatek za vsako vozlišče modela izračuna oceno napake diskretizacije. Dodatek sem tudi preizkusil na testnem primeru, ki je dal dobre rezultate.
Keywords:Metoda končnih elementov, prostorska diskretizacija, kvaliteta mreže, natančnost rezultatov, Abaqus, Richardsonova ekstrapolacija
Place of publishing:Maribor
Publisher:[J. Šantl]
Year of publishing:2015
PID:20.500.12556/DKUM-55455 New window
UDC:519.6:004.94(043.2)
COBISS.SI-ID:19407894 New window
NUK URN:URN:SI:UM:DK:YKC2OQYN
Publication date in DKUM:30.11.2015
Views:2384
Downloads:313
Metadata:XML DC-XML DC-RDF
Categories:KTFMB - FS
:
Copy citation
  
Average score:(0 votes)
Your score:Voting is allowed only for logged in users.
Share:Bookmark and Share



Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Secondary language

Language:English
Title:Analysis of FEM spatial discretisation influence on precision of numerical results
Abstract:Computer simulations are based on discretization (model is split into elements). This way, we obtain aproximated results of examinated problem. The quality of finite elements has a great effect on precision of numerical results. I described how the precision of numerical results is influenced by element type, integration scheme, geometrical shape and size. I described the problems that we face by using full or reduced integration scheme. The quality of the geometrical shape of finite elements is usually evaluated based on element aspect ratio, size of inner angles and Jacobian determinant. I researched the methods for evaluating discretizaztion error. One of them is Richardson extrapolation, which I researched in more detail. This method extrapolates observed value (like element stress), using observed values on three different meshes. This way, this method esimates the value of observed value for mesh with infinite small elements (the size of elements eqals 0). Based on this method, the mesh konvergence index (GCI) was introduced. GCI is used as a measure of discretization error. GCI defines an interval, which contains an exact solution of mathematical model with the probability of 95 %. Based on Richardson extrapolation and GCI index, I developed Abaqus add-on. The add-on calculates discretization error for each node of provided model, based on results on three different meshes. I tested add-on on test example, where I obtained good results.
Keywords:Finite element method, spatial discretization, mesh quality, result precision, Abaqus, Richardson extrapolation


Comments

Leave comment

You must log in to leave a comment.

Comments (0)
0 - 0 / 0
 
There are no comments!

Back
Logos of partners University of Maribor University of Ljubljana University of Primorska University of Nova Gorica