Can Edinburgh bond model in forum run without hertz contact?
Hi,
I have downloaded Timoshenko bond model from forum "https://support.altair.com/community?id=kb_article&sys_id=4bc876dd1b9620d0c4dfdbd9dc4bcbc8", but it seems that in this model when bond broken, particles contact with hertz contact, it will conflict with some basic model like HM with JKR. Can author suppport dll without hertz contact when bond broken like bondv2, thanks a lot!
Best regards!
Raheem
Answers
-
Hi,
The Timoshenko Beam Bond model was written with certain features in it that allow it to bond particles that have been created even if a large overlap exists. Internally it removes this overlap and continues the bonded calculation until bond failure occurs. When the failure happens the original large overlap comes into consideration for the non-bonded contact and for this reason, the non-bonded contact needs to remain built into the overall bonded model. We have looked at ways of separating this, but it would still depend on the other contact model being able to deal with the resetting of the intial bond time overlap which the current inbuilt contact model are not able to do.
The solution would be to integrate the desired contact model into the non-bonded portion alonside the current HM and LS models. While this is not a trivial task, it is possible. I don't see a simple way of setting the model to operate in a chain without losing the ability to deal with large initial bonding overlaps, which is quite useful.
May I ask what use case requires the bond and a cohesive contact? Does it specifically need to be an Adhesive contact model? Or do you just want to use some other cohesionless contact model such as linear spring?
Regards,
JP
0 -
J.P. Morrissey said:
Hi,
The Timoshenko Beam Bond model was written with certain features in it that allow it to bond particles that have been created even if a large overlap exists. Internally it removes this overlap and continues the bonded calculation until bond failure occurs. When the failure happens the original large overlap comes into consideration for the non-bonded contact and for this reason, the non-bonded contact needs to remain built into the overall bonded model. We have looked at ways of separating this, but it would still depend on the other contact model being able to deal with the resetting of the intial bond time overlap which the current inbuilt contact model are not able to do.
The solution would be to integrate the desired contact model into the non-bonded portion alonside the current HM and LS models. While this is not a trivial task, it is possible. I don't see a simple way of setting the model to operate in a chain without losing the ability to deal with large initial bonding overlaps, which is quite useful.
May I ask what use case requires the bond and a cohesive contact? Does it specifically need to be an Adhesive contact model? Or do you just want to use some other cohesionless contact model such as linear spring?
Regards,
JP
Hi JP
Thanks for reply, the material I want to simulate is a kind of fracture material cenmentated with Polymer with long length bond, it's the best if cohesion can be considered compared with the cohesionless case, but hearing from your explantation, it seems that the inbuilt models are not able to accomplish it.
Reagrds!
Raheem
0 -
J.P. Morrissey said:
Hi,
The Timoshenko Beam Bond model was written with certain features in it that allow it to bond particles that have been created even if a large overlap exists. Internally it removes this overlap and continues the bonded calculation until bond failure occurs. When the failure happens the original large overlap comes into consideration for the non-bonded contact and for this reason, the non-bonded contact needs to remain built into the overall bonded model. We have looked at ways of separating this, but it would still depend on the other contact model being able to deal with the resetting of the intial bond time overlap which the current inbuilt contact model are not able to do.
The solution would be to integrate the desired contact model into the non-bonded portion alonside the current HM and LS models. While this is not a trivial task, it is possible. I don't see a simple way of setting the model to operate in a chain without losing the ability to deal with large initial bonding overlaps, which is quite useful.
May I ask what use case requires the bond and a cohesive contact? Does it specifically need to be an Adhesive contact model? Or do you just want to use some other cohesionless contact model such as linear spring?
Regards,
JP
Hi,
Can this model generate bond when one type of bond already generated? In reference file the bond time was set for all type of particles, it seems that new generation of bonds will cove the old bond.
Regards!
Raheem
0 -
Raheem Sterling_22160 said:
Hi,
Can this model generate bond when one type of bond already generated? In reference file the bond time was set for all type of particles, it seems that new generation of bonds will cove the old bond.
Regards!
Raheem
Do you mean bonds created by another model such as the inbuilt EDEM model or bonds created by the TBBM?
In the case of the TBBM it should only created a new bond if the current contact is unbonded and the timestep is before the bond time specified in the preference file.
If you are seeing any other behaviour, let me know as that would not be the expected behaviour.
JP
0 -
J.P. Morrissey said:
Do you mean bonds created by another model such as the inbuilt EDEM model or bonds created by the TBBM?
In the case of the TBBM it should only created a new bond if the current contact is unbonded and the timestep is before the bond time specified in the preference file.
If you are seeing any other behaviour, let me know as that would not be the expected behaviour.
JP
Hi
Sorry, I didn't describe it clearly, for example, I have created ETBBM in my simulation with particleA and run the simulation with bond broken between particleA, and then, I added particleB and created ETBBM just between particleB, but it seems that the bond between particleA has been created again, because I didn't the broken bond between particleA.
In the custom property of ETBBM, I saw the "contact is bond" has value of 0 and 1. 0 means broken, I guess maybe you can add a value like bondstatus[] = 0 for non bonded, 1 for bonded, 2 for bond broken, then when I created new bond between particleB, the ETBBM between particleA with bondstatus[]=2 will not bonded again.
Raheem
0 -
Raheem Sterling_22160 said:
Hi
Sorry, I didn't describe it clearly, for example, I have created ETBBM in my simulation with particleA and run the simulation with bond broken between particleA, and then, I added particleB and created ETBBM just between particleB, but it seems that the bond between particleA has been created again, because I didn't the broken bond between particleA.
In the custom property of ETBBM, I saw the "contact is bond" has value of 0 and 1. 0 means broken, I guess maybe you can add a value like bondstatus[] = 0 for non bonded, 1 for bonded, 2 for bond broken, then when I created new bond between particleB, the ETBBM between particleA with bondstatus[]=2 will not bonded again.
Raheem
OK, yes, I see what you mean about if the bond has broken. You only want a bond to form if it has never previously been bonded.
I'll add this as a feature request and see if there is a simple way to keep both behaviours in the model.
JP
0 -
J.P. Morrissey said:
Hi,
The Timoshenko Beam Bond model was written with certain features in it that allow it to bond particles that have been created even if a large overlap exists. Internally it removes this overlap and continues the bonded calculation until bond failure occurs. When the failure happens the original large overlap comes into consideration for the non-bonded contact and for this reason, the non-bonded contact needs to remain built into the overall bonded model. We have looked at ways of separating this, but it would still depend on the other contact model being able to deal with the resetting of the intial bond time overlap which the current inbuilt contact model are not able to do.
The solution would be to integrate the desired contact model into the non-bonded portion alonside the current HM and LS models. While this is not a trivial task, it is possible. I don't see a simple way of setting the model to operate in a chain without losing the ability to deal with large initial bonding overlaps, which is quite useful.
May I ask what use case requires the bond and a cohesive contact? Does it specifically need to be an Adhesive contact model? Or do you just want to use some other cohesionless contact model such as linear spring?
Regards,
JP
Hi,
These days I have added a new bond model into EDEM, and want to bond particle with large overlap, as you said, when bond broken, the initial overlap was used for the force calculation, why not the current overlap was used for the calculation?
Reagrds
Raheem
0