Still pretty new: A few questions about modeling high plastic strain impact

[Luke720]

I hope this isn't too elementary for this forum but I was hoping some of you Abaqus gurus could give me a bit of guidance.

My goal is to model the impact of a spherical metal particle (40 micron diameter) onto a large (can be assumed to be semi-infinite) substrate at high velocities (750 m/s) to simulate the high strain and strain rate associated with the impact.

For my first version I am calling the substrate rigid and focusing on the deformation of the particle. I am planning to add levels of complexity to the model incrementally until it has the detail I want.

So far I have tried it two different ways and run into the same problem: It takes too long to submit. Additionally, this means I don't even know if it is working.

What I have tried:
1. A 3D sphere meshed with C3D10M elements. I figured that this may be unnecessarily intensive due the the symmetries in the scenario, so moving onto number 2...
2. I made an axisemetric model with about 1500 elements (tet at the center and quad radiating out), figuring that this would run MUCH faster. However I submitted it about 40 minutes ago and it is still running on a relatively powerful PC.

I have done some of the basic tutorials provided in abaqus and the submit step never took more that 30 seconds, so either I figure I must be doing something wrong. I also wonder if large plastic strains are having a detrimental effect on computation time.

I've attached the axisemetric attempt if that helps, let me know if the 3D one would be helpful to look at. Ideally I would like the later versions to be 2D or 3D so that I could model several particles impacting in succession at various locations.

Any help would be greatly appreciated!
-------------------------------------------------------------------
Additionally, if anybody has any advice on how to accelerate learning abaqus or methods that worked for you, I'd love to hear them. This simulation will be an integral part of my masters degree so I need to get good at abaqus. I know it's a slow process to learn software this complex and I have time, but the quicker the better as far as I am concerned.

Thanks,
Luke

 

[DanStro]

You are using a newer version of Abaqus than I so I can't open your file but here are couple of things to try.

- 40 minutes for a highly non-linear/short duration event doesn't sound unreasonable, at least in a general sense. The non-linearity isn't just because of the (possible) large deformation of your particle but also because of the contact conditions changing so rapidly.

- Try running the example "Oblique impact of a copper rod" or "Rigid projectile impacting eroding plate" to get an idea as to how long your system might take. These are 3D simulations so they should be closer to your #1.

- Make sure are you doing an Explicit analysis or Standard? It sounds like you should be doing an Explicit analysis.


HTH,
Dan

Han primo incensus

 

[Luke720]

Thanks Dan! I'll try running those this afternoon when I can get back to the computer lab.

I may need to coordinate with the computer lab to let my program run over night, because I am expecting very large deformation of the particle.

You mentioned contact which I didn't really mess with in my file. Do you think I have adjusted the contact parameters for this instance, and would that effect the computation time and the result?

I was running explicit, but it is good to hear that that is the way to go, I was not quite sure.

Luke

 

[Luke720]

I'm still having trouble with the same problem, I tried running it overnight and it still did not complete.

I also tried lowering the impact velocity to very small values so that the deformation would be very minor, it still didn't complete after a few hours.

If anybody has any other advice that would be greatly appreciated, this is driving me nuts.

 

[DanStro]

You had mentioned that you "...didn't really mess with..." contact. Does that meant that you aren't using contact or just that you haven't changed any of the default settings? You definitely need to have contact defined in the model that you described.

You also might want to look into using mass scaling. This can decrease the run time when running an explicit analysis. I can't be too much help with that since I've only used a couple of times.

Dan

Han primo incensus

 

[Luke720]

Thanks Dan, I didn't define any sort of contact (DOH!), I try that now.

I was under the impression that standard contact controls were build into the analysis when two separate parts came into contact. I'm going to do the contact turtorial now and try to implement what I learn there.

I will also look into mass scaling.

Again, thank you Dan, I know my issues may seem pretty simple but you have been a very helpful and I truly appreciate it.

 

[DanStro]

You need to tell Abaqus that you want it to consider contact conditions. Once you do a there are defaults that you can take advantage of.

dan

Han primo incensus