Stuck? 
support@mecway.com
support@mecway.com
Mecway
FEA
Victor
About
- Username
- Victor
- Joined
- Visits
- 4,282
- Last Active
- Roles
- Administrator
Comments
-
@disla, I seem to remember you found some problems with very high strains in linear materials before? Although that's usually physically unrealistic, in this case, the mechanical strain is near zero so I'd say it should be a reasonable approach. M…
-
Wow. I never knew there was any way to get rotational DOFs in equations! It's a bit of a pity that you have to keep one unused to put the dummy constraint on so it probably can't really be done automatically, but excellent discovery nonetheless!
-
While/after it's solving, click Details in the solver pane, then Open working directory and the .frd file is in there.
-
It might be too complicated. I'm trying to reduce it to the core problem (attached). I noticed that it does sort of solve if you add a "no extra contact step" in the CCX branch of the outline tree. The film is still flapping around but it's winding…
-
Which element it chooses isn't defined, sorry. It might be highest or lowest element numbers but might change in future. I do have to sort out that selecting hidden elements problem though.
-
-
You can put a dummy element like a zero-stiffness spring on that node so CCX includes it in the solution.
-
It seems like you can't have the master and slave directly opposite each other. My guess is it might see that as penetration and try to push them apart. All I can think of is to make 2 or 3 separate contacts in repeating stripes so the master for o…
-
It solved with the dent in the larger plate flattened out (set to y=0). I wonder if there's buckling happening around there leading to poor/no convergence. Is this something you're expecting or want to capture? With the dent in place, it solved a l…
-
Versions 24 and 25 install ccx/ccx_win64_mkl_pardiso_source_2.21.zip (numbers instead of xx) in the application's installation location. Compiling that produces both the ccx.exe distributed with that version of Mecway, and ccx_mkl.exe which is the s…
-
It's possible that turning off mixed precision (PASTIX_MIXED_PRECISION=0) takes away all the speed benefits of PastiX. I hadn't checked that. If you don't set it, it'll probably be faster but there's a risk of wrong results in some cases. A lot of …
-
It didn't finish solving in either 24 or 25 for me. It's possible you have a different version or build of CCX used by 24. 25 would have defaulted to the built-in one when it was installed.
-
That's probably the contact state file (.cel) which can be huge and slow to load. To prevent it being created, before solving, have only the field variables you want in the solution, and not contact state. If there's nothing there, Mecway automatic…
-
What I did: * Changed the tension force from nodes to faces to make it uniform. * Adjusted the X position of one side of the film's nodes so the elements are square, not parallelogram. This might not have been needed. * Ramped the tension for…
-
That's usually what causes it, but in this case, it's an unused node (876). Nodes without elements don't go through to the solution in CCX so delete it before solving.
-
CCX sometimes fails such large rotations at specific angles. You can work around it by altering the time step from 2 s to say 2.1 s and it seems to avoid whatever problem it got stuck on.
-
You can't have a time-dependent displacement on the same node as the time-dependent rotation. That's a limitation of Mecway. You also can't have constraints on the surface of node-surface coupling, which is a limitation of CCX. A workaround is to e…
-
You can use Node-surface coupling for that.
-
You're right, it doesn't have sufficient constraints. Beware that the pinned support only pins one point even if you apply it to two nodes. If you want two pinned supports, either use two of them or 3 displacement constraints on two nodes. There's…
-
If you're using large displacements, the line of action for the force from each screw might change as it bends, otherwise, for linear analysis, I think using the hole would be fine since you're interested in the behavior away from the hole. The sup…
-
I see. This isn't really consistent with an original design decision to keep node numbers hidden as much as possible and only using consecutive numbers. While it's possible, the ordering would be messed up if you delete something from the middle and…
-
I don't quite understand. But I'm trying not to have node numbers exposed as much as possible. You can change the node numbering of an element in Mesh tools -> Element properties, and there's Mesh tools -> Invert to swap to one other ordering…
-
It also works if you apply the force to the elements instead of their faces. I think the reason is that on faces, Mecway generates tiny moments due to numerical error and these might be what's breaking CCX.
-
It's supposed to work. This looks like a bug in CCX. It solves with CCX 2.16 but not 2.17-2.21 which have: *ERROR in add_sm_st: coefficient should be 0 It solves OK with quadratic elements so maybe you can use that instead? Generally a good idea a…
-
Thanks @disla for that very comprehensive exploration of ADJUST. Especially ADJUST=NO. I didn't know interference fit won't work without that! If I make this an input, would you say there's any value in allowing arbitrary distances or just ADJUST= …
-
@disla Good point. ADJUST is supposed to happen automatically for penetrating nodes which might explain why @JohnM's solution works. Somehow I thought it was already using ADJUST in all cases. When it's not, are the contact springs initially under …
-
CCX has viscoplasticity but it's not in the Mecway GUI. You can probably use it by writing some cards by hand. See *VISCO and *CREEP in the CCX manual. CCX also has other non-viscous material models. See *HYPERELASTIC and *HYPERFOAM.
-
I see. I've added it to my list but not sure when I'll get to it.
