Connect BEAM to TRUSS

Analysis issues
1

Using ccx 2.23. Similar to issue on Github.

inp
inp1373×1382 109 KB

Is it impossible to connect a beam element to a truss element?

  • Using a shared node does not work:
    The cross section of the left end of the beam should be rotated, but it isn’t.
    And the tangent to the beam on the right end should be horizontal, but it isn’t.

  • Using *EQUATION does not work either:
    The tangent to the beam on the right end should be horizontal, but it isn’t.

*NODE, NSET=ALL
1,    0,    0 ,0
2,    0, 1500, 0
3, 3000,    0, 0
**
** Using EQUATION:
777, 0, 0, 0
*ELEMENT,TYPE=T3D2,ELSET=ROD
1, 1,2
*ELEMENT,TYPE=B31,ELSET=BEAM
** 2, 1,3
** Using EQUATION:
2, 777,3
*MATERIAL,NAME=MAT
*ELASTIC
210.e3, 0.3
*SOLID SECTION,ELSET=ROD,MATERIAL=MAT
1000
*BEAM SECTION,ELSET=BEAM,MATERIAL=MAT,SECTION=RECT
200, 100
0, 1, 0
*BOUNDARY
2, 1,3
3, 1,6
**
** Using EQUATION:
*EQUATION
2
1, 1, 1.0,  777, 1, -1.0
2
1, 2, 1.0,  777, 2, -1.0
2
1, 3, 1.0,  777, 3, -1.0
*STEP
*STATIC
*CLOAD
1,2, -50.E3
*NODE PRINT, NSET=ALL
U, RF
*NODE FILE
U
*END STEP
2

Using a shared node:

shared
shared1800×1800 27.4 KB

3

Using *EQUATION:

equation
equation1800×1800 33.8 KB

4

I think the shared node problem is this issue:

I get a good solution with my own fork of CCX which has that bug fixed, but a fix isn’t in the official repo yet.

Could you post the actual model with the *EQUATION form? It’s not really clear what I should change to convert to that.

5

In my example, the deflection is the same for the shared node.

But the beam is not suffering from the loads it should be suffering from, I think.

If you want, you can share your results…

6

That bug changed over the versions. See the comments on it for the updates.

Here’s the solution with my fixed fork:

image
image646×430 3.48 KB

7

Yes, hmmmh, well, but the beam should have curvature!
The curvature should be zero on the left end and extreme on the right end.

And what about if you use the *EQUATION keyword instead of a shared node?

Well, anyway, let’s see, if the problem will be fixed.

8

It’s as curved as it can be with no midside nodes. Left and right end rotations look qualitatively consistent with the curvature they should have (none and hogging respectively).

Can you post the *EQUATION file? Maybe as code-formatted text directly in the message.

9 
*NODE, NSET=ALL
1,    0,    0 ,0
2,    0, 1500, 0
3, 3000,    0, 0
**
** Using EQUATION:
777, 0, 0, 0
*ELEMENT,TYPE=T3D2,ELSET=ROD
1, 1,2
*ELEMENT,TYPE=B31,ELSET=BEAM
** 2, 1,3
** Using EQUATION:
2, 777,3
*MATERIAL,NAME=MAT
*ELASTIC
210.e3, 0.3
*SOLID SECTION,ELSET=ROD,MATERIAL=MAT
1000
*BEAM SECTION,ELSET=BEAM,MATERIAL=MAT,SECTION=RECT
200, 100
0, 1, 0
*BOUNDARY
2, 1,3
3, 1,6
**
** Using EQUATION:
*EQUATION
2
1, 1, 1.0,  777, 1, -1.0
2
1, 2, 1.0,  777, 2, -1.0
2
1, 3, 1.0,  777, 3, -1.0
*STEP
*STATIC
*CLOAD
1,2, -50.E3
*NODE PRINT, NSET=ALL
U, RF
*NODE FILE
U
*END STEP

The derivation of a 2-node-beam-element also uses a nonlinear interpolation. So I expected this interpolation to also be used in the cgx visualisation.

But from your response I read, that cgx does not show this interpolation.

10

Thanks. I get the same problem as you with *EQUATION. I’d say it’s a new bug and deserves to go on Github.

Yea CGX is probably just plotting the nodes, not like, say Strand7 that can display 2-node beams with interpolated curvature.