Sorry if this has been asked before, but what is the best method to model bolt connections using beam elements? I have tried using node surface coupling on both of the beam ends, but the elements sheared apart.
I actually have the same question. So, I'm next in line after DCH.
Hoping to simulate the bolt spiders using node-surface coupling and inducing preload by node displacement. Preload by constraint equation would be nice, but it's not available in CCX.
Tested a preload idea by several runs of opposing displacements within a fixed ends column, but the Non-Linear CCX mode kind of of blew up and diverged for some reason I don't know (pic attached).
I've liked the 2-sec time step method Sergio and others have advocated with the thermal preload approach, and I wanted to apply that with the bolt spiders.
It's often easier to use solids for bolts instead of beams. You can connect the underside of the bolt head to the surface with bonded contact. If you make the solid mesh with a plane through the cross-section, you can also use CCX's *PRE-TENSION SECTION instead of thermal stress. *PRE-TENSION SECTION is also supposed to work on beams.
I had hoped the node-surface coupling would be useful for bolts but it doesn't work reliable when you connect an element to the reference node, so it's pretty much limited to remote force/mass/displacement or linkages.
@cwharpe, that's quite concerning. From the solid mesh deformed view, it looks like hourglassing.
I tried line3 and it does converge but to the wrong value, which is even worse! In that case, it looks like the displacement constraints aren't being applied to the midside nodes, allowing them to bulge outward and reduce the force.
What worked was changing the default line2 element type from B31 to B31R as in the attached file.
Modeling solid bolts is worth the effort. It's possible to model a fairly coarse solid bolt that behaves well.
Use the Library feature to keep pre-built bolt models handy. You can import, rotate, drag into position and stretch pretty easily. You can include the thermal stress trick, all of those features read in as well.
I have used the Calculix pretension feature on occasion to prove that it can work, but it takes some set up each time, and is direction dependent. This makes it less useful for plug-and-play.
I made this example awhile ago to answer a similar question. It shows how to model a bolt in the simplest way possible, using a solid model. I don't have any insight with trying to do something similar with beams.
[A belated Thanks to everyone who contributed an answer. Clearly, modeling a bolt with solids gives a superior result. Seems I learn more when things go wrong -- I'd never heard of hour-glassing before this.]
Lately I've been trying to verify the CCX Pre-tension examples. Finally realized they don't read-in to Mecway exactly -- seeing many red flags. Had better success by Importing, not Opening the example *.ini file, then saving under a new fname. When the new fname is opened, the warnings go away, but the CCX card (Pre-Tension in this case) must be re-pasted into CCX "custom model definition" before solving.
That being said, I'm frustrated over not being able to run example pret3, the one using the Beam element for Pre-Tension. I attempted to substitute Mecway Node-Surface coupling for CCX Distributed Coupling. Got as far as the green "Solver Exited" box. However, a Windows error box simultaneously flashed "Mecway has stopped working" and closed the program. Never any problems before with that message.
Has anyone successfully tried this example in Mecway?? Please share your methods.
Strange. I loaded the tutorial into Mecway, added the missing commands from the original inp file. If I run, Mecway will crash. But if I save the Mecway model as a CCX run file, and then submit directly to CCX, it works. (LEFT is ORIG CCX, RIGHT is Mecway-to-CCX) For some reason I had to scale the Mecway model by 1000x to get the units correctly, but that's it.
Import doesn't really work better than Open, it just doesn't show the error messages.
Sorry for the frustrating crash. I'm not sure quite why it happens but I'll fix it for the next version (14).
Here's my conversion of pret3.inp to .liml. Use Tools -> Options -> Calculix -> Working directorySame as the .liml file to make the .frd file easier to find. Then after it crashes, open the solution from pret3.frd.
*DISTRIBUTING COUPLING isn't the same as Mecway's node-surface coupling which uses the two separate keywords *COUPLING and *DISTRIBUTING. So I entered those cards manually.
As the title of the thread is Simple bolt modelling, I will throw in the use of a spring element. This gives a quick and easy fastener with preload capability. Useful if not concerned about bolt bending.
Comments
Hoping to simulate the bolt spiders using node-surface coupling and inducing preload by node displacement. Preload by constraint equation would be nice, but it's not available in CCX.
Tested a preload idea by several runs of opposing displacements within a fixed ends column, but the Non-Linear CCX mode kind of of blew up and diverged for some reason I don't know (pic attached).
I've liked the 2-sec time step method Sergio and others have advocated with the thermal preload approach, and I wanted to apply that with the bolt spiders.
~CW
I had hoped the node-surface coupling would be useful for bolts but it doesn't work reliable when you connect an element to the reference node, so it's pretty much limited to remote force/mass/displacement or linkages.
@cwharpe, that's quite concerning. From the solid mesh deformed view, it looks like hourglassing.
I tried line3 and it does converge but to the wrong value, which is even worse! In that case, it looks like the displacement constraints aren't being applied to the midside nodes, allowing them to bulge outward and reduce the force.
What worked was changing the default line2 element type from B31 to B31R as in the attached file.
Use the Library feature to keep pre-built bolt models handy. You can import, rotate, drag into position and stretch pretty easily. You can include the thermal stress trick, all of those features read in as well.
I have used the Calculix pretension feature on occasion to prove that it can work, but it takes some set up each time, and is direction dependent. This makes it less useful for plug-and-play.
I made this example awhile ago to answer a similar question. It shows how to model a bolt in the simplest way possible, using a solid model. I don't have any insight with trying to do something similar with beams.
http://mecway.com/forum/discussion/737/bolt-assembly-example#latest
anthony
Lately I've been trying to verify the CCX Pre-tension examples. Finally realized they don't read-in to Mecway exactly -- seeing many red flags. Had better success by Importing, not Opening the example *.ini file, then saving under a new fname. When the new fname is opened, the warnings go away, but the CCX card (Pre-Tension in this case) must be re-pasted into CCX "custom model definition" before solving.
That being said, I'm frustrated over not being able to run example pret3, the one using the Beam element for Pre-Tension. I attempted to substitute Mecway Node-Surface coupling for CCX Distributed Coupling. Got as far as the green "Solver Exited" box. However, a Windows error box simultaneously flashed "Mecway has stopped working" and closed the program. Never any problems before with that message.
Has anyone successfully tried this example in Mecway?? Please share your methods.
Strange. I loaded the tutorial into Mecway, added the missing commands from the original inp file. If I run, Mecway will crash. But if I save the Mecway model as a CCX run file, and then submit directly to CCX, it works. (LEFT is ORIG CCX, RIGHT is Mecway-to-CCX) For some reason I had to scale the Mecway model by 1000x to get the units correctly, but that's it.
Sorry for the frustrating crash. I'm not sure quite why it happens but I'll fix it for the next version (14).
Here's my conversion of pret3.inp to .liml. Use Tools -> Options -> Calculix -> Working directory Same as the .liml file to make the .frd file easier to find. Then after it crashes, open the solution from pret3.frd.
*DISTRIBUTING COUPLING isn't the same as Mecway's node-surface coupling which uses the two separate keywords *COUPLING and *DISTRIBUTING. So I entered those cards manually.
Also, had never expressly loaded an .frd file to check results. Until now. Nice workaround.