In the manual I saw the 1D element like coupling (distributed or kinematic )or dcoup3d and is mentioned how this element spread force Or element but no documentation are available about mass spreading.
Is there any suggestions/example/experience how to spread a 1D mass on a 3D component without adding extra stiffness?

Check this thread: *Coupling, distributing with beam elements - #14 by Disla

Please carefully study the many options available (*EQUATION, *RIGID BODY, *COUPLING, and DCOUP3Dâ€¦) because from my tests in the thread @Calc_em mentioned, I verified strange behaviors in some of them. It would be nice to know how 3D elements behave when concentrated mass is applied to them because I only verified it for simple 1D beams. The *inp files are attached in the other thread for reference.

Good luck!

I think you can put a mass element on each node with the values scaled in the same way surface traction would be. Calculating the individual mass values is more of a job for a preprocessor though unless your mesh is quite simple and has equally sized parallelogram or equilateral triangle faces. I tried it with a square face that has a mass of 1 spread over it and it looks like it could be correct but I havenâ€™t verified it properly.

2 Likes

Thanks to all I will try to put masses at element nodes scaled in according to distance from the reference node.

You want a non-uniform mass distribution? Even if you scale by distance from somewhere, you should also scale them according to contributing element face area (as traction would be) which I canâ€™t quite describe the details of.

Thanks I will do some check comparing results with other software.
The importante lesson learn that I understand is that for spreading a mass from a point to a component nodes we canâ€™t use 1D element in calclulix and we should apply the equivalent mass at element nodes in same way.
Do you agree?

Actually I might be misunderstanding you problem. Perhaps you want to connect a point mass to an area or volume so it doesnâ€™t apply a point load? In that case, you can use *RIGID BODY like @ lucas_bueno suggested but it makes the other part rigid though. There isnâ€™t an RBE3 type of â€śsoftâ€ť rigid body that works for mass. So yea I think you would have to apply radially decreasing point masses or write your own *EQUATIONs which would be a lot more work.

Abaqus has a special feature called â€śnon-structural massâ€ť for that. There are even multiple ways of defining it:

• distribution: mass proportional or volume proportional
• units: total, per volume, per area or per length

But itâ€™s also possible to attach point mass elements to reference nodes of coupling constraints. In CalculiX, that seems to be currently not supported (unless the workaround with `*EQUATION` works) so you are basically left with either rigid body constraint (at the cost of making a surface rigid) or manual attachment of properly distributed point mass elements to the nodes of the mesh.

1 Like

Some days ago, I comment about the option to artificially increase the density.
Isnâ€™t that Abaqus volume proportional distribution?

In case the density is increased artificially , one could benefit from the fact that density can be written as a function of temperature. That also allows to distribute mass according to a nonuniform temperature field previously defined.

Hereâ€™s how those 2 distribution types work:

• mass proportional (in proportion to the element structural mass) - uniform scaling of the structural density of the region, doesnâ€™t change the center of mass of the region
• volume proportional (in proportion to the element volume in the initial configuration) - uniform value added to the underlying structural density over the region, center of mass of the region may be altered if the region has nonuniform structural density

Of course, the mass per unit volume unit option is meant to add density.

Abaqus has mass scaling in explicit dynamics too but it serves a different purpose.

I would say that could be the triky solution.

I have add the mass by means of a temperature asigned to the beam node.
The density is temperature dependant . The difference is clear.
Density is internally correctly distributed in second order beam elements.

That confirms in some way there is a bug in Mass distribution for second order beam elements.

Yepâ€¦ Quite a nasty one. Mass, loads, BCsâ€¦ In all of them, the middle nodes are ignored.

Wouldnâ€™t that mean adding a negative mass to the corner nodes on of a second order hex element?

@rsmith Corner nodes, yea. It still seems to work.

Coupling type Distributing work for load type to be transfered, itâ€™s eliminates overstiffening the models. Unfortunately, not for the case of mass, only type Kinematic seems to be work but still have overstiffenng as Rigid body.