Does small elements attached to RB decrease Timestep?

[fcolomb 0]

Hello,

 

I am running an explicit simulation and I have a forming tool that due to its geometry requires a small mesh, much smaller than the part of interest. Since I don't want to investigate anything that happens with the tool, all its elements are linked to a Rigid body. My "element check tool" is showing that the timestep for these elements will be really small. Will the solver use this related timestep as the minimum of my run?

 

I don't think it will be counted, but just being sure, because I couldn't find this information elsewhere.

 

Thank you.

[Guest]

Hi,

Two-time step methods are available, 

 

Element time step

Nodal time step

 

It is possible that timestep is controlled by a node even while using elementary timestep due to contacts. 

 

The timestep used by the solver is the largest available time step.

If the element time step is used, element size affects the time step. 

 

Use nodal time step in general to save computation time.

 

[fcolomb 0]

Hello @Prakash, thank you for the reply.

 

Yes, I am aware of the two kinds of time step and I know the courant condition that dictate the time step too. But my doubt is for example whether elements attached to a Rigidbody that is totally constrained, will also play a hole on this count or no. Because since they won't have any stress/strain, there should be no calculation on them. But I don't know if the software "thinks" like that.

 

If the solver "choose" this time step as the one to be used, can I reduce the young modulus of these elements (the ones attached to the RB) to reduce their timestep? I know is not good to add mass to rbody, but what about reducing stiffness on it's slave elements, since I don't care about them?

[Guest]

If the timestep of the element is larger then the same will be used by the solver. Again, you can confirm this from out file and take necessary action. 

 

1 hour ago, fcolomb said:

If the solver "choose" this time step as the one to be used, can I reduce the young modulus of these elements (the ones attached to the RB) to reduce their timestep? 

Yes, bu this may affect the overall result. 

[Hyperman 293]

Hi @fcolomb,

 

nodes constrained by the rigids will not decrease the timestep during simulation. You can check this simply by deleting the rigids and comparing the timestep in the outfile.

 

It is not a good idea to change stiffness or mass of the elements constrained by rigids, because this will change the physics of the problem. The stiffness is especially important for contacts as both master and slave stiffness can be taken into account for contact stiffness, depending on the Istf flag.

[Guest]

13 minutes ago, Ivan said:

nodes constrained by the rigids will not decrease the timestep during simulation.

Yes, but the master node of the RBODY can give time step. 

 

15 minutes ago, Ivan said:

It is not a good idea to change stiffness or mass of the elements constrained by rigids, because this will change the physics of the problem. The stiffness is especially important for contacts as both master and slave stiffness can be taken into account for contact stiffness, depending on the Istf flag.

By adding mass we are already changing the stiffness which eventually affects the results. But limiting this mass addition is acceptable say 2%...?!!

[Hyperman 293]

@Prakash Pagadala

 

can you explain which factors affect the master node timestep of he RBODY?

[Guest]

Inertia, mass on RBODY

[fcolomb 0]

Thank you very much the answers guys.

 

Yes @Ivan, that was exactly my doubt. Now I understand it, its good to know that I can use small elements attached to Rbody without increasing running time. And also, very useful information about the relation between stiffness and contact, I wasn't aware of that.

 

About adding mass to the Rbody, if I am using impdisp on this specific Rbody the whole time, does the mass on it affect time step too? I mean, the inertia effect are controlled by my impdisp, right?