Timoshenko Beam Bonded Particle Model is not supported by GPU CUDA Solver?

Question

Hello,
I downloaded the tutorial and example of a uniaxial crush test of bonded particles using the Timoshenko Beam Bonded Particle Model. However, I tried to run with the GPU CUDA Solver, but EDEM showed an error message. Is this model only available on CPU?

Thanks.

Link source: https://community.altair.com/community/en/edem-api-contact-model-example-timoshenko-beam-bonded-particle-model?sys_kb_id=f73ef0c4dba6e910cfd5f6a4e296197a&id=kb_article_view&sysparm_rank=4&sysparm_tsqueryId=e3b07325dbba39d4cfd5f6a4e29619ea

Julia Bonaldo · Answer

Hello,
Thank you again.
In this case in the factory do I need to set the bond creation time to the current time or fixed=0s?
Can you please check if the steps I intend to do are correct?

1) I created a particle factory with a bond creation time to the current time, but I added only two contact models: Hertz with JKR and the standard rolling friction.

2) At the beginning of the compaction (end of factory) I removed the Hertz model, added the EEPA, and kept the standard rolling friction.

3) The sample is compacted and unloaded keeping the EEPA and the standard rolling friction.

4) I will export the deck to a time=0s and include the EBPM model (keeping the EEPA and rolling friction as well). The 'BondedParameters' txt file (see attached) I added to the same path where it's installed the model library EBPM.dll: C:\Program Files\Altair\2023\EDEM\lib\ContactModels (I read this in your Reference Manual in PDF). In the  'BondedParameters.txt' I included several lines because I have 5 different particles size. I will not add the EBPM model for the interaction between particle and geometry.

P.s: The reason why I decided to not include any bond model since the beginning (compaction) is that I realized that the calculation time becomes much slower when including bonds, and the EEPA model works fine for the compaction and unloading, so the bonds will not help in my case.

Thanks!

BondedParameters.txt

jpmorr · Answer

You could actually do either option since the TBBM supports a non-zero bond time, but I usually find that it's easier to prepare a deck and then export to a new one with a start time of zero.

Then add the TBBM in the new deck to replace the other contact models beofre bonding.

JP

Julia Bonaldo · Answer

Thank you, Morrissey.
If I need to prepare the sample by compaction and unloading before doing a crush test, for example, do I need to apply the EBPM since the particles factory or I can use another contact model for the factory, compaction, and unloading and then export the simulation and create bonds based using the EBPM to do the crush test?
For the factory, compaction, and unloading I am using the Edinburgh elastoplastic model, the bonding V2 (which in this case I would not use anymore if I apply the EBPM I guess), and standard rolling friction.

Thanks!