[rsarma]

Hi everyone,

I am writing a udf for a dynamic mesh update in parallel solver. I have attached a part of my code. Basically I use a looping macro, that loops across all the faces, then loops the nodes of that face and updates the nodal positions.

#if !PARALLEL
begin_f_loop(f,thread)
{
if PRINCIPAL_FACE_P(f,thread)
{
f_node_loop(f,thread,n)
{
v = F_NODE(f,thread,n);
if (NODE_POS_NEED_UPDATE(v))
{
NODE_POS_UPDATED(v);
NODE_X(v) = NODE_X(v) + deltax;
}
}
}
}
end_f_loop(f,thread)


The problem is when I use this is parallel, it does not work properly. Basically what it does is, in a parallel solver, the mesh is partitioned along cells. So there are certain nodes which belong to two (or more) faces (or cells). These nodes are updated twice (or more) independently by each of the compute nodes to which they are assigned. The macro (NODE_POS_NEED_UPDATE) does not seem to work in parallel case. Is my observation correct? Could you suggest any other macro that takes care of these shared nodes in a parallel solver? Please let me know your suggestions.

Best,
R

[Grigor Nikolov]

commands such as if !RP_HOST or if !RP_NODE have to be used to avoid this problem

[nausheensultana]

Hi R,

Did you find a solution to the problem? I have similar problems with node count, where the nodes are being counted more than once.

[AlexanderZ]

Originally Posted by nausheensultana

Hi R,

Did you find a solution to the problem? I have similar problems with node count, where the nodes are being counted more than once.

regarding script from first message, you may try to use

begin_f_loop_int

instead of previous macro

begin_f_loop

you may find more information in ansys fluent customization manual

best regards