|
[Sponsors] |
August 29, 2010, 20:22 |
Determinant 3x3x3
|
#1 |
New Member
Przemek
Join Date: Feb 2010
Location: Warsaw Poland
Posts: 27
Rep Power: 16 |
Hi,
could You explain me the definition of the Determinant 3x3x3 quality parameter? Beceause, the one from ICEM documentation isn't clear for me. Thanks, Batis |
|
November 16, 2010, 07:47 |
|
#2 |
Senior Member
Join Date: Mar 2009
Posts: 138
Rep Power: 17 |
Hey Batis,
The Determinant3x3x3 is supposed to be higher than 0.2; if you have negative values your volumes are supposed to be negative. The Determinant(3x3x3) is defined as the Determinant of the Jacobian Matrix... But I don´t know what this means in detail. I would also be gratefull if somebody could clarify it. - How can I get from my cell to a matrix? - What does it mean for my cell if I have low values (between 0 and 0.2)? Because just from seeing the cells with low Det. I can´t notice anything bad. thanks Camoesas |
|
December 5, 2010, 20:33 |
Determinant
|
#3 |
Senior Member
Simon Pereira
Join Date: Mar 2009
Location: Ann Arbor, MI
Posts: 2,663
Blog Entries: 1
Rep Power: 47 |
Hey all, I will put the official Help for determinant below, but basically, 2X2X2 is for a linear hexa element (2 end points on each edge). The 3X3X3 is for quadratic Hexas (3 points on each edge). You will usually get the same answer for both checks because you have linear elements. The 2X2X2 is faster, so just use that.
Greater than 0.2 is great. But most solvers will handle greater than 0.05 or 0.01. Determinant (2x2x2 stencil) The Determinant, more properly defined as the relative determinant, is the ratio of the smallest determinant of the Jacobian matrix divided by the largest determinant of the Jacobian matrix. In this option, the determinant at each corner of the hexahedron is found. The default range is 0–1 with a Determinant value of 1 indicating a perfectly regular mesh element and 0 indicating an element degenerate in one or more edges. Negative values indicate inverted elements. Determinant (3x3x3 stencil) This is for hexahedral elements. This option is the same as the 2x2x2 stencil, but edge midpoints of blocks are added to the Jacobian computation. The Jacobian determinants for hexahedrals will be calculated at r,s,t = -1,0,1 of the natural coordinate system of the element (27 node positions). Next it calculates the maximum absolute determinant of the 27 determinants (3x3x3). If this is at position i with absolute determinant value max0, then for each of the 27 positions (j) (except i) the absolute distance of determinant j to determinant i will be calculated. The final result will then be 1 minus the maximum of the absolute distances divided by max0, so that the range of this quality criterion value will be between -1 and 1. The Jacobian determinant is the determinant of the Jacobian operator which connects the derivatives of the natural coordinates (r,s,t) with the derivatives of the local coordinates (x,y,z). J = ((dx/dr dy/dr dz/dr) (dx/ds dy/ds dz/ds) (dx/dt dy/dt dz/dt)). -------------------------------------------------------------------------------- Note: A good book to understand the determinant calculation is: Finite Element Procedures, by K.J. Bathe, Prentice Hall, New Jersey 07632, 1996. |
|
April 30, 2018, 07:57 |
mesh and determinant
|
#4 |
New Member
José Ramón Quiñonez Arce
Join Date: Oct 2017
Location: Leeds, UK
Posts: 17
Rep Power: 9 |
Hi all,
I am trying to do a mesh for a combustor by using ICEM and the solver CFX, however my determinant I think is too bad. Nevertheless,f I refine it more in some corners, the determinant gives me negative results. I have tried a lot of different ways to do the mesh and this is the one gives me "better results", can you please help or give me some advise? which other aspects do I have to consider? I attach dome pictures below Many thanks!! mesh 1.jpg mesh 2.jpg mesh 3.jpg |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
ICEM11 get mesh -ve volume and -ve determinant | cfx2008 | CFX | 1 | October 28, 2008 16:44 |
Determinant in ICEM-CFD and Fluent | Anand | Main CFD Forum | 2 | August 18, 2001 00:53 |