• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 7
  • 1
  • 1
  • Tagged with
  • 22
  • 18
  • 15
  • 12
  • 11
  • 10
  • 8
  • 6
  • 6
  • 5
  • 5
  • 4
  • 4
  • 3
  • 3
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
21

Interval Matching and Control for Hexahedral Mesh Generation of Swept Volumes

Shepherd, Jason F. 01 April 1999 (has links) (PDF)
Surface meshing algorithms require certain relationships among the number of intervals on the curves that bound the surface. Assigning the number of intervals to all of the curves in the model such that all relationships are satisfied is called interval assignment. Volume meshing algorithms also require certain relationships among the numbers of intervals on each of the curves on the volume. These relationships are not always captured by surface meshing requirements. This thesis presents a news technique for automatically identifying volume constraints. In this technique, volume constraints are grouped with surface constraints and are solved simultaneously. A sweepable volume has source, target and linking surfaces. The technique described in this thesis uses graph algorithms to identify independent, parallel sets of linking surfaces, and determine if they correspond to through-holes or blind-holes. For blind-holes, the algorithm generates constraints that prevent the hole from being too deep in interval parameter space and, thus, penetrating opposite target surfaces. For each linking set, the adjoining source and target surfaces are partially ordered by the structure of the linking set. A small set of representative paths for each linking set is found, and the representative paths for all linking sets are gathered and distilled by Gaussian elimination into a small set of constraints.
22

Mesh discretization for modelling continuous casting in CFD : A comparison of tetrahedral, polyhedral and hexahedral mesh elements when modelling turbulent flow / Mesh discretization for modelling continuous casting in CFD : A comparison of tetrahedral, polyhedral and hexahedral mesh elements when modelling turbulent flow

Hjeltström, Johanna January 2024 (has links)
Continuous casting is by far the most widely used casting method for steel production all over the world. The Swedish steel industry has specialized in niche products with high quality. To obtain high-quality steel, flow control is crucial when operating the caster. Computational fluid dynamics is used in the research and development of steel production to gain knowledge and optimize flow control, with meshing being a crucial step. One challenge with meshing is finding the balance between computational cost (time and storage) and accuracy. This study investigated whether polyhedral mesh elements would be suffcient for modeling turbulent flow in continuous casting. Firstly, Ansys Fluent's polyhedral converter was used on an existing mesh containing tetrahedral and hexahedral elements. The mesh quality in terms of skewness, orthogonality, and aspect ratio was studied. The results showed that the polyhedral converter lowered the cell count and slightly improved the quality compared to the original mesh. However, more research is needed with other meshes and actual simulations to determine if it is suffcient for modeling turbulent flow in continuous casting. Following that, two new meshes were generated: one polyhedral and one poly-hexcore mesh, both using Ansys Fluent Meshing, and their mesh quality was analyzed as in the first case. These meshes were then simulated with the Large Eddy Simulation model for steel, slag, and air for 60 seconds. They were compared and evaluated using production data in the form of steel layer fluctuations and nail board measurements. The results from the polyhedral and poly-hexcore meshes showed that both meshes had good, equal quality and cell count. Even so, due to the construction of the poly-hexcore, switching between polyhedral and hexahedral elements, and the occurrence of hanging nodes, the simulation showed signs of discretization errors. This might have also contributed to the lower flow velocities at the steel layer compared to the nail boards for the poly-hexcore mesh. The polyhedral mesh showed more stable simulation results with closer velocity trends compared to the nailboard measurements. The steel layer fluctuations for both meshes were somewhat similar to the production data, but further FFT analysis concluded that fluid flow simulations alone are not enough for an FFT comparison to production data since many factors are missing from the simulations, such as mould oscillations and regulations from the stopper. The computational cost varied between time and storage, with the polyhedral mesh requiring 13 % less simulation time but 32.7 % more storage. In summary, it can be concluded that polyhedral mesh elements can be successfully implemented for modeling turbulent flow in continuous castingwith a reasonable computational cost. / Stränggjutning är den främsta metoden använt för stålproduktion över hela världen. Den svenska stålindustrin har specialiserat sig på nischade produkter med hög kvalitet. För att uppnå högkvalitativt stål är flödeskontroll avgörande vid drift av gjutmaskinen. Numeriska strömningsberäkningar används i forskningen och utvecklingen av stålproduktion för att skaffa kunskap och optimera flödeskontrollen, där meshing är ett avgörande steg. En utmaning med meshing är att hitta balansen mellan beräkningskostnad (tid och lagring) och noggrannhet. Denna studie undersökte om polyhedrala mesh-element skulle vara tillräckliga för att modellera turbulent flöde i stränggjutning. Först användes Ansys Fluents polyhedrala konverterare på en befintlig mesh, innehållande tetrahedrala och hexahedrala element. Meshkvalitén i termer av skevhet, ortogonalitet och aspektförhållande studerades. Resultaten visade att den polyhedrala konverteraren sänkte cellantalet och förbättrade kvaliteten något jämfört med den ursprungliga meshen. Dock behövs mer forskning med andra mesher, och faktiska simuleringar, för att avgöra om den är tillräcklig för att modellera turbulent flöde i stränggjutning. Därefter genererades två nya mesher: ett polyhedralt och en poly-hexcore, båda med hjälp av Ansys Fluent Meshing, och precis som i det första fallet, studerades deras meshkvalitet. Dessa två meshes simulerades sedan med Large Eddy Simulation-modellen för stål, slagg och luft i 60 sekunder. De jämfördes och utvärderades med hjälp av produktionsdata i form av stålnivå, fluktuationer och 'nail board'-mätningar. Resultaten från det polyhedrala och poly-hexcore meshen visade att båda mesherna hade god och likvärdiga kvalitéer samt cellantal. Trots detta, på grund av uppbyggnaden av poly-hexcore meshen, alltså växlingen mellan polyhedrala och hexahedrala element och förekomsten av hängande noder, visade simuleringen tecken på diskretiseringsfel. Detta kan också ha bidragit till de lägre flödeshastigheterna vid stålnivån jämfört med 'nail board'-mätningarna för poly-hexcore meshen. Den polyhedrala meshen visade mer stabila simuleringsresultat med hastighetstrender närmare 'nail board'-mätningarna. Amplituden för stålnivåfluktuationerna för båda näten var något lika produktionsdatan, men en vidare FFT-analys visade att endast flödesimuleringar inte är tillräckliga för en FFT-jämförelse med produktionsdata eftersom många faktorer saknas i simuleringarna, såsom svängningar av gjutformen och reglering av inloppashastigheten. Beräkningskostnaden varierade mellan tid och lagring, där den polyhedrala meshen krävde 13 % mindre simuleringstid men 32,7 % mer lagring. Sammanfattningsvis kan det konstateras att polyhedrala mesh-element framgångsrikt kan implementeras för att modellera turbulent flöde i stränggjutning med en rimlig beräkningskostnad.

Page generated in 0.032 seconds