Apply moment to 1d beam element

[tomato]

I have a cantilever beam to which I’d like to apply moment

I’ve tried using *MPC MEANROT and modelling the moment “as is”: creating two “ghost” points to which I applied *CLOAD. However, none produced any result. What am I doing wrong?

The screenshots of results can be viewed here: Applying moment - Album on Imgur

mpc_meanrot.inp

*Node, NSET=Nall
1, 0, 0, 0
2, 2000, 0, 0
3, 200, 0, 0
4, 400, 0, 0
5, 600, 0, 0
6, 800, 0, 0
7, 1000, 0, 0
8, 1200, 0, 0
9, 1400, 0, 0
10, 1600, 0, 0
11, 1800, 0, 0
12, 100, 0, 0
13, 300, 0, 0
14, 500, 0, 0
15, 700, 0, 0
16, 900, 0, 0
17, 1100, 0, 0
18, 1300, 0, 0
19, 1500, 0, 0
20, 1700, 0, 0
21, 1900, 0, 0
22, 0, 1, 0

*Element, TYPE=B32R, ELSET=Eall
1, 1, 12, 3
2, 3, 13, 4
3, 4, 14, 5
4, 5, 15, 6
5, 6, 16, 7
6, 7, 17, 8
7, 8, 18, 9
8, 9, 19, 10
9, 10, 20, 11
10, 11, 21, 2

*ELSET,ELSET=M0B0RstdD0
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,

*MATERIAL, NAME=MechanicalMaterial
*ELASTIC
210000, 0.3

*BEAM GENERAL SECTION, ELSET=M0B0RstdD0, MATERIAL=MechanicalMaterial, SECTION=PIPE
50, 5
-0, 1, 0

*MPC
MEANROT, 2, 2, 2, 22

*STEP
*STATIC

*BOUNDARY
1, 1, 6

*BOUNDARY
22, 1, 1, 0.9

*NODE FILE, OUTPUT=3d
U
*EL FILE
S, E

*NODE PRINT, NSET=ConstraintFixed, TOTALS=ONLY
RF

*END STEP 

mpc_beam.inp

*Node, NSET=Nall
1, 0, 0, 0
2, 2000, 0, 0
3, 200, 0, 0
4, 400, 0, 0
5, 600, 0, 0
6, 800, 0, 0
7, 1000, 0, 0
8, 1200, 0, 0
9, 1400, 0, 0
10, 1600, 0, 0
11, 1800, 0, 0
12, 100, 0, 0
13, 300, 0, 0
14, 500, 0, 0
15, 700, 0, 0
16, 900, 0, 0
17, 1100, 0, 0
18, 1300, 0, 0
19, 1500, 0, 0
20, 1700, 0, 0
21, 1900, 0, 0
22, 2000, 0, 100
23, 2000, 0, -100

*Element, TYPE=B32R, ELSET=Eall
1, 1, 12, 3
2, 3, 13, 4
3, 4, 14, 5
4, 5, 15, 6
5, 6, 16, 7
6, 7, 17, 8
7, 8, 18, 9
8, 9, 19, 10
9, 10, 20, 11
10, 11, 21, 2

*ELSET,ELSET=M0B0RstdD0
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,

*MATERIAL, NAME=MechanicalMaterial
*ELASTIC
210000, 0.3

*BEAM GENERAL SECTION, ELSET=M0B0RstdD0, MATERIAL=MechanicalMaterial, SECTION=PIPE
50, 5
-0, 1, 0

*MPC
BEAM, 2, 22

*MPC
BEAM, 2, 23

*STEP
*STATIC

*BOUNDARY
1, 1, 6

*CLOAD
22, 1, 100

*CLOAD
23, 1, -100

*NODE FILE, OUTPUT=3d
U
*EL FILE
S, E
*NODE PRINT, NSET=ConstraintFixed, TOTALS=ONLY
RF

*END STEP 

[Calc_em]

But beam elements have rotational DOFs even in CalculiX so you can apply the rotation directly to the beam end node.

[tomato]

You mean something like this: (2 is the beam end node)

*BOUNDARY
2, 5, , 0.01

Then, 0.01 is radians, right? How can I “convert” it to torque? (Or torque to radians)
With “ghost” nodes and applied force to them I know how to calculate torque: perpendicular_distance(ghost_node, beam_end) * force_applied_to_ghost_node, but I don’t understand how to achieve the same through rotation

[Calc_em]

Yes and you can also use *CLOAD in the same way:

*CLOAD
2, 5, 1000

then it will be torque instead of force (1000 Nmm here - just some dummy value).

[vicmw]

Although this example is OK, beware that the moment or rotation should be aligned with the beam’s axis and/or the beam should be aligned with a global axis otherwise it gives the wrong result.

The problem occurs with all ways - *BOUNDARY, *CLOAD, and *MPC since all of them use the mean rotation MPC internally.

Wrong displacement for beam with a single rotational DOF constraint.

[xyont]

indeed, rotational transmit at beam end node from boundary or load seems to have a problem in some case. If i can remember correctly, several early version of CalculiX generate knot at these nodes to eliminate. However, later versions change the behavior by mean rotation. Maybe a new specific keyword like *Knot is required to assign manually of rigid section surfaces at location interest.

[vicmw]

Yea I would love to be able to force a knot at a node, and don’t know why Guido changed it to avoid knots? A workaround is to attach a dummy beam hanging off the node with different orientation or section but that’s a bit ugly looking.

[xyont]

below my approach to get identical results, and it seems not too bad in looking.

[vicmw]

Replace one beam with two half-width beams? That’s an idea. It’ll make the solution look a bit weird with all those extra nodes and elements though, so it would be nice to have an invisible way.

[xyont]

this approach is similar to composite shell available with some exception and limitation.