Minimum Bolt size is 28 elements and 189 Nodes.
First *2 refinement leads to 224 Elements and 1173 Nodes.
How do you suggest to connect this bolt to the plates? I would like to do some comparisión.
Sometimes even hybrid solid-beam-solid models are used, this could help with potential issue with too few elements in contact (but of course, may need quite a lot of effort to define):
(wanderlodgegurus.com/database/Theory/FEA%20Bolt%20and%20Bolted%20Joint%20Modeling.pdf - btw. it’s a really interesting document showing various approaches)
in case of solid model and beam element for bolts, simplification by three beams per bolt as figure below. There’s possible setup for prestressed and tension only.
Congratulations 
. Seems like you’re starting to understand the proposal you’ve criticized so much. Although only halfway.
This is a comparision between your superior aproach and the simplified proposed one for this particular loading condition with shear and bending involved.
¿?¿? I think you still don’t know how to evaluate shear in a bolted connection.
So wider beams for head and nut as proposed before, but connected via rigid body constraints ? And the connection at the transition level between both beam sizes is done to the middle layer of nodes within the same plate. The question is how beam elements are connected with rigid body constraint reference nodes.
no any enlarging beam diameter to mimic’s bolt head or nut, surface tied or penalty contact in modeling. Additional beam required to connect mid surface of plate connected, also another MPC is used instead of rigid body.
this modeling required feature of draw 1D element by two nodes or multiple (master and slave nodes) like spider element in PrePoMax. It could be better to post new discussion in CalculiX forums since related to solver capability and limitation.
Ok, I see. Did you compare with a single beam approach (connected only to surfaces under head and nut) ? What MPCs were used ? The symbols belong to rigid bodies, but they were removed from the model ?
Right, we are talking about approaches requiring extensive keyword edits and abstracting from PrePoMax anyway, so indeed it would indeed be better to move to the CalculiX forum.
1 Like
I’m currently trying to model the proposed solution using a B32R beam and distributing couplings via the following function. However, I’m getting no convergence.
*Coupling,REF NODE=93003,SURFACE=sHead1,CONSTRAINT NAME = CPBOLT1_HEAD
*DISTIBUTING
1,6
When I remove the contact between my two parts A and B, the results show that my upper part is not coupled to my two bolts.
Would you happen to have an example model showing how to apply distributing couplings?
It got lost in this overly long thread, but there’s an official example from the CalculiX test suite here: Modeling bolt connection with preloaded beam element - #5 by FEAnalyst
Try adding equation constraints. The approach with distributing couplings can be problematic, though.
1 Like
test_bolt_v6.pmx (4.3 MB)
You may have to use *DISTRIBUTING COUPLING instead. Apart from the example linked above, see this video (link to file download is provided in the description): https://www.youtube.com/watch?v=AiLCzKAOdx0
Differences between these coupling types are covered here: Different coupling constraints and their limitations - CalculiX (official versions are on www.calculix.de, the official GitHub repository is at https://github.com/Dhondtguido/CalculiX).
Something like this (just check if I got the node numbers right):
test_bolt_v6 mod.pmx (8.0 MB)
i do simply connect beam element to references nodes and results shown bending and shear being transferred. Still prefer another MPC by general equation constraint to make it sure, this modeling approach of beam connected to rigid body is not allowed in latest CalculiX version also.
to get higher accuracy, it’s required gap element, nonlinear spring and large equation in modeling of bolt simplification as my previous schematic figure.
1 Like
Not sure what you are doing but can’t be right if both bolts show up the same stress values.
that’s shame even by simple and clearly explanation still hard to understand. Result is reasonable at least instead OP following one modeling approach and yields to zero value or unstressed bolts no matter loads direction and magnitude assigned.
The node numbers are correct and the nodes are properly coupled. However, the model shows very large displacements and the analysis does not converge.
I have reviewed the video you suggested as well as the different examples provided, but I still cannot understand the behavior of my model.
test_bolt_v7.pmx (9.2 MB)
The connection is hinged causing rigid body motion of the top plate. Keep in mind that *DISTRIBUTING COUPLING in CalculiX can’t handle rotations/moments. But even *KINEMATIC COUPLING will have the same problem. Try with 3 beams per bolt like in the synt’s screenshot above. You should also activate contact between the plates and then try to get it to converge (maybe with displacement controlled loading for now).
1 Like
Would it be possible to obtain your .pmx file?
Because, as shown in the table, its reference node doesn’t handle displacement BCs at all. Kinematic coupling would make more sense.