Global moment equilibrium issue with TIE connections between thickened shell meshes v2

[slave.k]

Hello everyone,

As a continuation of my battle with welded steel structure described here:

Global moment equilibrium issue with TIE connections between thickened shell meshes

Hello everyone, I would like to share an observation regarding global moment equilibrium in models created using the “Thicken Shell Mesh” function in PrePoMax 3.0. While working on a relatively complex welded steel structure (approx. 60 TIEs), I noticed that the global force equilibrium was practically perfect, whereas the moment equilibrium showed significant discrepancies. After systematically removing components and constraints, I narrowed the problem down to *TIE connections between plate…

I have found additional behavior leading to global moment equilibrum discrepancy.

Prepomax 3.0 + Calculix 2.22

Surface model with Thickened Shell Mesh - C3D20, 80x40x4mm

Three plates connected to each other by TIE Constraint

Vertical plates are moved 1mm out of base plate

There are 2 seperate tie constraints, but master is same surface in both cases, position tolerance 1mm, Adjust No

Loads - 2 separate Surface Traction, each 100N in every global direction, BC (0;0;0mm) on sidewall of horizontal plate, as shown above.

Result

Looks correct but analysis of force and moment (in rel. to 0;0;0) equilibrum shows:

                       FORCE [N]                     MOMENT [Nm]

RF: [-200.000065, -200.000081, -200.000007] [-12.542234, -4.259411, +20.400007]
CLOAD: [+200.000000, +200.000000, +200.000000] [+16.200000, +4.200000, -20.400000]
GRAV: [+0.000000, +0.000000, +0.000000] [+0.000000, +0.000000, +0.000000]
DIFF: [-0.000065, -0.000081, -0.000007] [+3.657766, -0.059411, +0.000007]

When it is connected only one vertical plate (no matter which) to horizontal plate, moment equilibrum is OK.

I noticed also, that when there is only one plate connected, in analysis log there is statement like this:

*WARNING in gentiedmpc: no tied MPC
generated for node 36122
no opposite master face
found; in-face tolerance: 1.4240350858498002E-002
*WARNING in gentiedmpc: no tied MPC
generated for node 37120
no opposite master face
found; in-face tolerance: 1.4240350858498002E-002
*WARNING in gentiedmpc: no tied MPC
generated for node 37465
no opposite master face
found; in-face tolerance: 1.4240350858498002E-002

what is logical and related to 3 nodes which are out of master surface. Strange thing is that when there are 2 plates connected to horizontal one, there is no such warning at all.

Solution

Separating master for right and left plate solves the issue.

Swapping master and slaves also solves issue. In that case max displacement is 1,86mm vs. 1,67 as above

I can send .pmx file if needed.

Questions:

In my complete welded structure I have many big surfaces connected to small edges of other plates (area relation higher than 200:1). What is more correct/better: define Tie master surface on big area and slave on small or opposite, or it doesn’t matter? I know other rules about meshing, stiffness.

[Calc_em]

@slave.k

I can send .pmx file if needed.

Could you share the .pmx and possibly also .inp file here using some hosting service ?

@slave.k

Separating master for right and left plate solves the issue.

It’s a good practice to limit the constraints and contact to minimum surfaces, but such issues may still be unexpected and require further investigation. When you share the files, I can compare it with Abaqus results.

[slave.k]

Files are in:

Moment_equilibrum

[Calc_em]

Here’s what I get in Abaqus:

• forces (red) and moments (blue):

• visualization of tie constraints:

• displacements:

[Stefano4000]

Calculated here with 2.22,
better is here basically actual version 2.23 to go not in solved problems.
Is here a fast calculation here possible?
Same results in 2.22 and 2.23?