simulating rock impact on steel plate using MES

[tim]

Greetings,

I have been using ASM 2014 to model impact of a rock (similar to granite) on a steel plate structure.  I am using MES for this.  As well as stress and deflection, I am interested in obtaining the impact force.

The model has been running reliably so no problems there.  The results appear generally as you would expect.

I have been setting the element definition to isotropic  for both rock and steel.  I intend to change the steel to a non linear von mises model to allow for yielding becausae the stress results are well into the yielding regime.

I have used mostly 8 node bricks with some 20 node bricks where necessary.

I am using a surface to surface contact.

The rock is modelled as a sphere.  I have selected granite as a material.

To obtain impact forces, I have used the x and y reactions -> inquire results->sum technique. (first I selected the nodes in the contact region of the rock).  I then combined the x and y reactions to get a resultant reaction as the plate is inclined to the x-axis)

The forces and stresses however, seem higher than expected. 

I have 5 basic questions:

1.  What should i set the rock element definition to?  It is highly likely that the rock will fracture near the point of impact.

2.  What sort of accuracy can I expect wrt stress and impact forces?

3.  Is the technique for impact force calculation correct?  I am wondering about the correct selection of the nodes.

4.  Are there any good references for modelling impacts of this type?

5.  Are there any ways to check these results?

Regards,

Tim

[AstroJohnPE]

Hi Tim,

Here are my opinions to your 5 questions.

1. Rock element definition? May not matter because finding material properties for the "more exotic" material models is almost impossible. Also, the software will not show the rock breaking apart; at best, it will indicate that the rock "should have broken". If you want to explore other material models, I think (but do not know for sure) that Drucker Prager material model would work. There is also a concrete material model, but I do not remember if it is for small displacement only, or if it would work in your large displacement analysis.
2. The overall stresses should be accurate (+/- 10% in ideal model?) but the local stress at the point of impact will not be accurate. I think the impact force obtained like you are doing (summing the reactions) is not accurate. Ideally, the impact force would converge to "the answer" as you make the timestep smaller and smaller. I think you will find that the impact force increases indefinitely as the time step is made smaller. I do not know why this happens.
3. Yes, your technique is correct. In fact, if you select a few nodes away from the impact site, you will see that the reaction is 0. So there is some leeway with what nodes you select. Obviously, you do not want to accidentally select nodes with constraints since they will have a reaction, too.
4. I do not know.
5. It should not be too hard to find a rock and steel plate that you can bang together. 😉 Even if you do a simple drop test and measure the reaction force, this would give you some data points to compare with the simulations.

Good luck. Let us know how it turns out.

[tim]

John,

Thanks very much. Yes, as I made the timestep smaller, the impact force increases! Maybe a hotfix sometime! Impact is a big deal with our mining industry, so it would be good to improve our modelling technique and software to cope with this.

I will try to validate this sometime, somehow. In Australia, our miners should be able to find a spare rock and steel plate. A set of kitchen scales and a quick eye ought to get the job done.

Actually, a simple test like that is measureable if I replace the scales with a small flexible beam or plate. Then all I have to do is measure the beam deflection after impact with plasticine under the beam or something similar.

Regards,

Tim