Hello,
I'm trying to perform an explicit analysis with mecway.
I know that is necessary use Calculix keyword , I've used DYNAMIC,EXPLICIT but is the right way?
I've add this keyword to custom model definition but run fail.
Some one could provide a right way to use ccx explicit solver and\or a simple example? (crashing tube or other simple example)
Thanks
Comments
that seem to be implicit dynamic
there is a way to create a mecway model with contact+plastic material behavior+dynamic like in other explicit solver, for example like in Radioss?
Add DYNAMIC,EXPLICT to custom model definition is not the right way?
Thanks
the way to call explicit solver from mecway gui directly
Could you explain me how use gui for this?
This example doesn't solve very well and I don't know why. Maybe the time step size is wrong.
Also beware that plastic and contact don't work together with nonlinear dynamic. This seems to be true of explicit dynamics too, so that could be a showstopper.
thanks for your support.
Sad too read that plastic and contacts don't work thogether.
You know if this is a bug that could be fixed in the next release of calculix?
Thanks again
This is a test problem that is very "fragile". In other words, it can crash pretty easily.
Some things I know -
- Use auto-time stepping with no lower bound but restrict the upper time step (it will probably be a smaller value than you want).
- Softer contact stiffness helps
- I use "modify keyword" but you will likely have to put command in custom model/custom step definition to put all of the switches and things like initial velocities.
- use initial condition to set velocity vs dropping through a large distance
- suppress all stresses while debugging, files get very big, and i/o will slow things down
- use "save every 10th (or 50th)" option in Analysis Options once things look good
Things I don't know-
Everything else. I don't know how any of the *DYNAMIC options or damping setting really work, but I think this just takes some time to work out.
When you learn how use the options for *DYNAMIC and Rayleigh damping, do share
Here's a video of the dropped box:
https://drive.google.com/file/d/1AcXEopF_XPwfg98zRBd_fZDvQ8wTZpan/view?usp=sharing
Thanks for sharing
I should study more deeper how solver work.
I've notice that shell elements works not well with explicit (only error for now but could be a problem of my setting)
I hope that next releases of calculix improve explicit solver
Thanks at all
but I belive that an explicit code should manage contacts+plastic+dynamics...soon or late
I also agree with you
I have been able to make some humble progress with Explicit. At least I feel more confident to make it converge. Take all what I will say carefully.
Explicit Dynamics account for quickly changing conditions and discontinuous events. Selecting the right time step and accommodating the model known that value is at this moment my main procedure.
This is so far where I have arrived and how I have proceeded. I’m still working on it.
1-Determine the maximum allowed time step . (Search for Volumetric COURANT information in the ccx progress window).
The stiffer your material is, the faster the waves propagate in your material and the shorter must be your time step to properly capture the impact.
We are talking of timesteps in the order of 1e-6 / 1e-8 s for steels !!!
If this increment time is not respected the nodes overlap and elemental volumes become negative. That effect is noticeable in my model. Imagine the mesh as a lot of springs connected to each other. Your time step must be small enough to capture the propagating wave from one node to the next.
2-I would forget about the pendulum. Gravity gives problem and known the time increment required and initial speed you will easily realize your problem would take a long time to impact. I have set up the impact as one dimensional with the hammer as close as possible to the target. You don’t want to spend computational time to see how a hammer advance. You want to see the impact. Recall I have set up a time step no bigger than the initial stable time increment.
3-You can establish the desired energy of the impact or impose potential energy = kinetic energy to obtain the initial velocity at the contact instant.
I’m adding this to the hammer (*) but I have also added it to the target (velocity =0) so all the model has initial conditions.
4-Linear elements. Quadratic shows massive distortion (kind of hourglass).
5-I have read Damping reduces the initial stable time increment. Nonlinear Dynamic only allows Damping proportional to the inertia. It doesn’t have too much sense to me yet as it doesn’t allow to remove the high frequencies (those due to internal wave propagation , precisely the ones that affect my convergence). Haven’t solve this yet.
6-Increase number of allowable increments
7-I have had to accommodate the contact parameter according to the stiffness of the materials involved in the impact.
At this point I need to find how to damp internal waves propagation and adjust courant to avoid crushing the elements.
I'm solving with Pardiso. Let me know if you get any additional progress.
(*)-25
What would be interesting is how to split up the critical time stepping. For example, with the pendulum impact problem, the early motion is easy, it's when the pendulum impacts that you need very fine time steps. I have not figured out a way to tell CCX how to do this, so I am restricted to the smallest step you need.
When you don't need impact, the solver does a really neat job on dynamics. Check out this video, this was run in CCX.
https://youtube.com/shorts/rXhOUTOTNsM?feature=share
. Thank you Sir.
¿Could the pendulum be a rigid body?
¿Why two spheres? Their contact is difficult by itself. Maybe better flat surface against your target?. (Floor is flat)
I’m slowly understanding better how explicit works.
High frequency excitations are controlled through a different parameter on the dynamic card.
it's time to give the ball a good kick.