DCOUP3D difficulties with temperature

2

Not sure if that’s right. Degrees of freedom should be natural numbers ¿isn’t it?. Now there is a point after the 1.

*DISTRIBUTING COUPLING,ELSET=Rbe3_Elem_1
Rbe3_circle, 1.

You have two DCOUP elements with the same ref node. I haven’t try this before.

*ELEMENT,TYPE=DCOUP3D,ELSET=Rbe3_Elem_1
** synthax: DCOUP3D element numb - node numb
30000, 80000
*DISTRIBUTING COUPLING,ELSET=Rbe3_Elem_1
Rbe3_circle, 1.
*ELEMENT,TYPE=DCOUP3D,ELSET=Rbe3_Elem_2
30001, 80000
*DISTRIBUTING COUPLING,ELSET=Rbe3_Elem_2
Rbe3_Plate, 1

Your seems unconstrained in both displacements and temperatures.
You need the temperature initial condition. The expansion is computed from a AT.
I would also assign it to the DCOUP3D node.

I would cut the model to a few elements untill you solve the issue.

3

From my understanding, this is the (weight) value => a real value accordingly.

Yes i’m trying to “force” the 2 parts to remain “coaxial”: I’m trying to mimic the Rbe3 elements behavior:

  • master nodes: screw hole ones
  • slave node: screw one
    The slave node motion is the average of the master ones: one DCOUP3D for the screw and one for the hole => they’re supposed to remain coaxial even if they expand (my assumption).

see:

*BOUNDARY
80000, 1, 3, 0.
80000, 11, 11, 600
4

Sure. Sorry, I had Distributing and Kinematic in mind. I don’t use Dcoup often.

It 's rotationally unconsrained.

imagen
imagen855×504 28.3 KB

There is no limintation, at least with Static (both with and without NLGEOM )

That looks like the initial condition is missing to reduce the number of unknows. There is a initial temperature needed as reference to start.

*INITIAL CONDITIONS,TYPE=TEMPERATURE
All_nodes,273.15

5

Using *INITIAL CONDITION doesn’t change anything (0°C is the default value anyway).

Node 80000 is fixed (Ux, Uy, Ux are nuls - we’re connected to 3D elements); the goal is to let the parts freely expand.

To better understand, just replace the DCOUP3D elements by *RIGID BODY ones … but the Rbe2 elements cannot expand with the temperature.

I’m wondering if “network equation” is linked with *NETWORK MPC? (but seems to be for fluid elements ???)

6

Does your model work without DCOUP3D?

7

The DCOUP3D elements have been replaced by Rigid Body ones (Rbe2); the applied temperature increase has been limited to 30°C (delta T = 10°C) to avoid overstress concentration and convergence issues … since Rbe2 cannot “expand” here.

disp
disp1109×592 62 KB

But is what you were speaking about?

8

VM
VM1090×596 80 KB

9

I see. Mine also fails with DCOUP3D and *Uncoupled/Coupled temperature-displacement.

10

Try completing your Specific Heat material definition with some value. At least the problem is now convergence.

*Material, Name=Steel
*Elastic
210000, 0.3
*Expansion, Zero=20
1.6E-05
*DENSITY
1.0E9
*CONDUCTIVITY,TYPE=ISO
300.
*SPECIFIC HEAT
20.
** Sections ++++++++++++++++++++++++++++++++++++++++++++++++
**
*Solid section, Elset=Circle, Material=Steel
*Solid section, Elset=Plate, Material=Steel
**

**
*Step, Nlgeom, Inc=10
*Uncoupled temperature-displacement, Solver=Pardiso, Steady state
1.0, 1.0, 1E-05,1E30
**
** Output frequency ++++++++++++++++++++++++++++++++++++++++
**
*Output, Frequency=1
**
** Boundary conditions +++++++++++++++++++++++++++++++++++++
**
*Boundary, op=New
** Name: Temperature_BC
*Boundary
All_nodes, 11, 11, 300
*BOUNDARY
80000, 1, 3, 0.
80000, 11, 11, 300
**
** Loads +++++++++++++++++++++++++++++++++++++++++++++++++++
**
*Cload, op=New
*Dload, op=New
*Cflux, op=New
**
** Defined fields ++++++++++++++++++++++++++++++++++++++++++
**
**
** History outputs +++++++++++++++++++++++++++++++++++++++++
**
*Node print, Nset=Spy_node, Global=Yes
NT
**
** Field outputs +++++++++++++++++++++++++++++++++++++++++++
**
*Node file
RF, U, NT, RFL
*El file
S, E, HFL, NOE
**
** End step ++++++++++++++++++++++++++++++++++++++++++++++++
**
*End step