hexahedra elements & accuracy???????

May 9, 2006, 10:13

Hello friends. I have a question. Is there any relation between the type of element used and the accuracy?? I am using CFX software. In the help files, it has not been given. Please help me. Regards, VVJ.

ericnutsch · December 3, 2009, 13:01

Serval professors have informed me that a Hexahedral mesh is more acurate in the boundary layer. Tetrahedrals have less skew computational issues and are better over irregular geometry or when expanding the size of the cells away from the geometry(size function).

I am curious however as to how much more accurate they are; since in the opensource world tetrahedrals are much easier to generate.

(Anyone) Do you think it would take twice as many
tetrahedrals, three times as many, four, or more to be equivalent to
the quality of a hexahedral boundary layer mesh?

CapSizer · December 3, 2009, 16:12

Well, in real life this is actually a very complex question with no simple answer. There are theoretical methods that should give you the relative accuracies, but in my experience practical concerns tend to override these. For example, it can depend on the topology of whatever it is that you are meshing. When it comes to something very regular like a 3D wing, for example, a tetrahedral grid may require 10 times as many cells as a hexa grid, for equivalent accuracy.

It is also noticeable how fast certain types of unstructured hexa grids will converge, by virtue of minimal skewness and good cell aspect ratios. I have sometimes seen a poor quality hexa grid produce better results than a good tetra grid. It rather depends on where the bad cells are, and how resistant the solver is to being tripped up by these bad cells.

kingjewel1 · December 3, 2009, 16:32

Can I refer you to this paper by Hefny and OOka: CFD analysis of pollutant dispersion around buildings: effect of cell geometry, Building and Environment 44 (2009)

ericnutsch · December 4, 2009, 14:37

Thanks CapSizer! You are the first person to put forth a number. 10 times means a whole lot more than "more". Im not going to base any calcultions on it, but it is good to have a feeling for the numbers before you choose a direction.

Thanks for the ref kingjewel1. Here is the abstract if anyone else is interested:
http://cat.inist.fr/?aModele=afficheN&cpsidt=21358896

If anyone else has had different experiences please post them. Thanks!

May 9, 2006, 10:13	hexahedra elements & accuracy???????	#1
vijesh joshi Guest Posts: n/a	Hello friends. I have a question. Is there any relation between the type of element used and the accuracy?? I am using CFX software. In the help files, it has not been given. Please help me. Regards, VVJ.

December 3, 2009, 13:01	hexahedral vs tetrahedral	#2
ericnutsch Senior Member Eric Nutsch Join Date: Sep 2009 Location: Eugene, Oregon USA Posts: 113 Rep Power: 16	Serval professors have informed me that a Hexahedral mesh is more acurate in the boundary layer. Tetrahedrals have less skew computational issues and are better over irregular geometry or when expanding the size of the cells away from the geometry(size function). I am curious however as to how much more accurate they are; since in the opensource world tetrahedrals are much easier to generate. (Anyone) Do you think it would take twice as many tetrahedrals, three times as many, four, or more to be equivalent to the quality of a hexahedral boundary layer mesh?

December 4, 2009, 14:37	hexahedral mesh is _x as acurate	#5
ericnutsch Senior Member Eric Nutsch Join Date: Sep 2009 Location: Eugene, Oregon USA Posts: 113 Rep Power: 16	Thanks CapSizer! You are the first person to put forth a number. 10 times means a whole lot more than "more". Im not going to base any calcultions on it, but it is good to have a feeling for the numbers before you choose a direction. Thanks for the ref kingjewel1. Here is the abstract if anyone else is interested: http://cat.inist.fr/?aModele=afficheN&cpsidt=21358896 If anyone else has had different experiences please post them. Thanks!

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Temporal accuracy	ganesh	Main CFD Forum	0	February 29, 2008 05:32
Hexahedran elements and accuracy????	vijesh joshi	CFX	0	May 9, 2006 10:14
Penetrating elements in extruded mesh	Michael P	CFX	2	May 20, 2005 08:06
accuracy of Discontinuous Galerkin	Qu Kun	Main CFD Forum	0	September 6, 2004 09:19
CFX4.3 -build analysis form	Chie Min	CFX	5	July 12, 2001 23:19

December 3, 2009, 16:12		#3
CapSizer Senior Member Charles Join Date: Apr 2009 Posts: 185 Rep Power: 18	Well, in real life this is actually a very complex question with no simple answer. There are theoretical methods that should give you the relative accuracies, but in my experience practical concerns tend to override these. For example, it can depend on the topology of whatever it is that you are meshing. When it comes to something very regular like a 3D wing, for example, a tetrahedral grid may require 10 times as many cells as a hexa grid, for equivalent accuracy. It is also noticeable how fast certain types of unstructured hexa grids will converge, by virtue of minimal skewness and good cell aspect ratios. I have sometimes seen a poor quality hexa grid produce better results than a good tetra grid. It rather depends on where the bad cells are, and how resistant the solver is to being tripped up by these bad cells.

December 3, 2009, 16:32		#4
kingjewel1 Senior Member Join Date: Jul 2009 Posts: 260 Rep Power: 17	Can I refer you to this paper by Hefny and OOka: CFD analysis of pollutant dispersion around buildings: effect of cell geometry, Building and Environment 44 (2009)