Oka wear model - Wrong behaviour when using deformed geometries
Hello,
I am working on a wear simulation of tillage tools. I need a particle bed for the process simulation. The particle bed is supported at the bottom by a geometry element. At the sides, however, periodic boundaries are used for support.
Now to my actual problem:
The geometry element at the bottom should not actually participate in the wear process, as it only serves to hold the particles in position above it.
Initially, I only defined interactions between the particles and the tool material (wear constants, hardness, ...).
Accordingly, no interaction parameters were set for the base plate material.
Nevertheless, the geometry of the base plate wears very heavily (or even disappears completely in some cases) from the first contact with particles. Even if I define interaction parameters for the base plate or change the mesh fineness, this can only help to a limited extent.
In comparison, the Archard Wear model works correctly.
Is the problem with the Oka Wear model known or is there a solution?
I have attached an adapted simulation file which shows the problem.
Best regards
HE
Best Answers
-
Hi, @H_Engbert could you try the updated API model on the community:
This should resolve the problem you are seeing if you use the API version rather than the built-in one.
Regards
Stephen
0 -
Hi, it shouldn't matter the order that the wear models are set, so long as they are after the Hertz-Mindlin model as the contact forces get passed down the chain. Both models use the HM force but don't modify the calculated value.
0
Altair EmployeeAnswers
-
Hi,
I do see the error, I'll report it to the product team to review.
It looks like it only occurs on GPU, our API version of the OKA wear model doesn't currently support GPU otherwise that could be an option, I'll check if anybody can update it and we can see if that can be used as a workaround.
RegardsStephen
0 -
Hi Steven,
Thank you for your quick response.
Just to make sure there are no ambiguities:
The .dll and .txt files included with the model are for a modified bonding model and accordingly do not belong to Oka Wear model. The wear model was simply selected from the physics models already included in EDEM and should be a standard model, shouldn't it?Best Regards
HE
0 -
Hi,
Thanks for the clarification, i saw that the custom bond model was included. I modified the test case so it was just a few particles with no bond model and still saw the issue so I've reported that to the product team.
Although it happens with the internal Oka model we do have an API version of this, which is the same physics just an external version. Unfortunately this doesn't currently support CUDA GPU as it's an older implementation. Once updated to CUDA this API version should resolve the issue, as I would expect the problem is related to some non-initialised values in the code rather than the specific physics, and should help as a shorter term fix rather than waiting for an updated version of EDEM.
1 -
Hi, @H_Engbert could you try the updated API model on the community:
This should resolve the problem you are seeing if you use the API version rather than the built-in one.
Regards
Stephen
0 -
Hi @Stephen Cole,
i've tested your updated API Modell for Oka Wear. It seems like it's working correct.
I really appreciate your fast solution to this problem.
One last question: When using Archard Wear and Oka Wear together in a single Simulation for modeling abrasvie and erosive wear, which Modell should be the first in the Physics tree?
0 -
Hi, it shouldn't matter the order that the wear models are set, so long as they are after the Hertz-Mindlin model as the contact forces get passed down the chain. Both models use the HM force but don't modify the calculated value.
0
