# Thread: Fortran progam which can generate constraints automatically for Abaqus

1. Junior Member
Join Date
2012-03
Posts
10

## Fortran progam which can generate constraints automatically for Abaqus

Hello, everybody:
I'm learning applying periodic boundary conditions in Abaqus and I've found it's very difficult to define *Equation constraints directly in Abaqus because there are so many nodes on opposite faces. One may develop a Fortran program which can generate all constraints automatically. Could someone who has this kind of experience give me an example? Maybe Abaqus has provided users some good way, if so, please tell me.
I think that will be a big favor for me, thanks!

2. Member
Join Date
2006-09
Location
Sweden
Posts
81
Make a nested (two-level) loop that loops over all surface nodes and identify those who are offset by the size of the representative volume element (i.e. those that are periodic). If your RVE is 1x1x1 you should look for nodes that are at distances 1, sqrt(2) and sqrt(3).

Mats

3. Junior Member
Join Date
2012-03
Posts
10
But why should I look for nodes are at distances sqrt(2) and sqrt(3)?Could you explain it to me in more detail?
Originally Posted by matsgd
Make a nested (two-level) loop that loops over all surface nodes and identify those who are offset by the size of the representative volume element (i.e. those that are periodic). If your RVE is 1x1x1 you should look for nodes that are at distances 1, sqrt(2) and sqrt(3).

Mats

4. Member
Join Date
2006-09
Location
Sweden
Posts
81
Well, I am just thinking that if your RVE is a unit cube, then there are three "types" of periodic node pairs:
1) Nodes that are internal to the cube surfaces,
2) nodes that lie on the edges of the cube, but not at the corners,
3) nodes on the corners of the cube.

In general, nodes on the edges have several periodic "buddies", at distances of 1 or sqrt(2). For the corners, the distances would be 1, sqrt(2) or sqrt(3).

(Sqrt(3) is the diagonal of the cube.)

Mats

5. Junior Member
Join Date
2012-03
Posts
10
But I think if I define all the nodes that are at distance 1, then the node that are at distances sqrt(2) and sqrt(3) will satisfy periodicity automatically.
Originally Posted by matsgd
Well, I am just thinking that if your RVE is a unit cube, then there are three "types" of periodic node pairs:
1) Nodes that are internal to the cube surfaces,
2) nodes that lie on the edges of the cube, but not at the corners,
3) nodes on the corners of the cube.

In general, nodes on the edges have several periodic "buddies", at distances of 1 or sqrt(2). For the corners, the distances would be 1, sqrt(2) or sqrt(3).

(Sqrt(3) is the diagonal of the cube.)

Mats

6. Member
Join Date
2006-09
Location
Sweden
Posts
81
Yeah, perhaps. I suppose the requirement is that each node on the surface of the RVE should be involved in a constraint equation (one per dof).
Mats

7. Junior Member
Join Date
2012-03
Posts
10
I see.Thanks for explaining these to me!
Originally Posted by matsgd
Yeah, perhaps. I suppose the requirement is that each node on the surface of the RVE should be involved in a constraint equation (one per dof).
Mats