the LQ(5,*) and LQ(6,*) maybe in the other's direc
 

In the posdat.f, you can use common to get the IC of the neighbor cell through the variable LQ(1:6,). Normally, the star-cd thinks LQ(5,*), and LQ(6,*) are the x direction's neighbors.

However, sometimes,in some cases, the LQ(5,*) and LQ(6,*) maybe in the other's directions. and all the directions are messed up.

Can anybody explain this? Thanks.

Re: the LQ(5,*) and LQ(6,*) maybe in the other's d
 

I can help you with this. check your mail.

Murali

Re: the LQ(5,*) and LQ(6,*) maybe in the other's d
 

<font face = "courier new"> Your qn:

My problem is like this:

LQ(1,IC): the -z direction neighbor cell;
LQ(2,IC): the z direction neighbor cell;
LQ(3,IC): the -y direction neighbor cell;
LQ(4,IC): the y direction neighbor cell;
LQ(5,IC): the -x direction neighbor cell;
LQ(6,IC): the x direction neighbor cell;

This is the rule in the star-cd user subroutine;

However, in some cases, I found that

LQ(1,IC): the -x direction neighbor cell;
LQ(2,IC): the x direction neighbor cell;
LQ(3,IC): the -y direction neighbor cell;
LQ(4,IC): the y direction neighbor cell;
LQ(5,IC): the -z direction neighbor cell;
LQ(6,IC): the z direction neighbor cell;



Answer:

Your assumption that 1,2 and 3 correspond to x,y and z directions is wrong. STAR-CD does NOT recognize x,y and z directions but only arbitrary directions.

1 - Bottom
2 - top
3 - South
4 - North
5 - West
6 - east

It is a li'l difficult to understand what these mean. See the following pic:


----|-----|-------
| N |
----|-----|-------
W | Cell| E
----|-----|-------
| S |
----|-----|-------
The cell in the middle is flanked by 4 cells in 2-d and 6 cells in 3-d. In 2-d u see North, East, West and South neighbours only and in 3-D, u see a top and bottom neighbour also. These are NOT related to x,y,z directions. They just give you the various cells attached through the face indices running from 1 to 6.

Then how can we determine a particular x, or y or z neighbour ?

(1) checking the centroidal co-ordinates. Suppose it is a well-ordered grid. If LQ(k,IP) returns a cell, you can access the centroidal information of this cell using the commoned variable CX. If the z-coordinate is greater than the cell's coordinate, it is a "positive" z-beighbour else it could be negative. (ofcourse, this may not be true if the hexahedral cells are distorted).

(2) Finding the normal of the faces using some other commoned variables e.g. NDIN. If the normal is same as z vector, your neighbouring cell is a z neighbour.

Actually I find this a little too difficult when compared to the neatly structured way of accessing information of neighbouring cells in FLUENT 6. But you have no other choice.

Regards

Murali