Bond failure due to particle seperation

Matthias Vink
Matthias Vink Altair Community Member
edited August 9 in Community Q&A

Hi,

I am currently trying to simulate a very flexible material in EDEM using the bondingV2 model. However, to achieve sufficient flexibility I had to use a low bond stiffness. But that caused the bonds to break due to particle seperation. Now I could increase the contact radii, but then bonds form over other particles, and not just with nearest neighbours. which then results in a rigid truss like structure. How can I fix this?

Matthias

Tagged:

Answers

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited March 21

    Hi Matthias,

    Have you considered / explored disabling the torque feedback option?  By disabling that no tangential torque will be applied between the bonds, making your bonds VERY flexible.
    image

    Let me know if that solves your problem!

    Eric

  • Matthias Vink
    Matthias Vink Altair Community Member
    edited March 21
    Eric Veikle_22252 said:

    Hi Matthias,

    Have you considered / explored disabling the torque feedback option?  By disabling that no tangential torque will be applied between the bonds, making your bonds VERY flexible.
    image

    Let me know if that solves your problem!

    Eric

    Hi Eric, yes I tried that, it does provide flexibility, however, when using a more complex shape (a chicken fillet) with quite a few sub particles. A sort off truss-like structure forms:image

    Which has the result that the structure is still rigid, to combat this stiffness values can be set very low but then the original problem arises. Is there a setting I can change that allows for more stretching of the bonds or perhaps a different bonding model exists that is more useful?

    Matthias

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited March 21

    Hi Matthias,

    As  far as the bond model, I think you are using the best one - Bonding V2.

    Two other suggestions that might be worth investigating.  First, you could layering your structure with two different materials, thereby allowing you to increase your contact radius beyond the particle physical radius, but not bond the neighbor 's neighbor.  Not an ideal solution because it adds complexity to the setup of your meta-particle and will increase the number of contacts that the model needs to address.  

    image

    Alternatively, you could try to adjust the contact radius after the bonds have been formed.  I believe that the contact radius is not stored and can be changed after the creation of the particles.  That would prevent you from needing to re-redefine the meta-particle, but would still come with the drawback of more contacts to resolve.

    Can you share more about your application, what physical thing are you attempting to simulate?

    Eric

     

  • Matthias Vink
    Matthias Vink Altair Community Member
    edited March 22
    Eric Veikle_22252 said:

    Hi Matthias,

    As  far as the bond model, I think you are using the best one - Bonding V2.

    Two other suggestions that might be worth investigating.  First, you could layering your structure with two different materials, thereby allowing you to increase your contact radius beyond the particle physical radius, but not bond the neighbor 's neighbor.  Not an ideal solution because it adds complexity to the setup of your meta-particle and will increase the number of contacts that the model needs to address.  

    image

    Alternatively, you could try to adjust the contact radius after the bonds have been formed.  I believe that the contact radius is not stored and can be changed after the creation of the particles.  That would prevent you from needing to re-redefine the meta-particle, but would still come with the drawback of more contacts to resolve.

    Can you share more about your application, what physical thing are you attempting to simulate?

    Eric

     

    Hi Eric,

    Thank you for your suggestions! I am trying to model a (3D) chicken fillet that both represents the bulk and single behaviour of real fillets. 

    How do I exactly re-define the contact radius after the bonds have been formed? I can imagine this will be complicated when generating fillets with a dynamic factory.

     

    Matthias

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited March 22

    Matthias,

    That is an interesting application, what is the aim of the simulation?

    My suggestion of changing contact radius would only be valid if you generated all the fillets before changing the radius, if you need a dynamic factory it will not work.

    Given that you need to use a dynamic factory my suggestion of layering you fillets with different types of particles is the only way that I can think of to allow for larger contact radii.

    Without knowing what your goal is with the simulation I am not sure if this suggestion is valid, but you could partition your fillet into rigid sections and bond the rigid sections together using the bond model. In this case I would use the multi-sphere builder to generate the sphere filled sections of the fillet, and place them together with a Meta-particle for bonding.
    Kind Regards,

    Eric

  • Matthias Vink
    Matthias Vink Altair Community Member
    edited March 25

    The aim is to streamline automation in the poultry industry. Creating a (good enough) model of chicken fillets will allow for the judgement of concept viability without having to build prototypes (and thus waste resources).

    I think there is potential in increasing the contact radius after meta-particle placement. How do I do that? I tried modifying the contact radius in the creator section in a timestep where the meta-particle is already created and then going back to the simulator tab and continue simulating but the fillet still fell apart.

    Kind regards,

    Matthias

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited March 25
    Matthias Vink said:

    The aim is to streamline automation in the poultry industry. Creating a (good enough) model of chicken fillets will allow for the judgement of concept viability without having to build prototypes (and thus waste resources).

    I think there is potential in increasing the contact radius after meta-particle placement. How do I do that? I tried modifying the contact radius in the creator section in a timestep where the meta-particle is already created and then going back to the simulator tab and continue simulating but the fillet still fell apart.

    Kind regards,

    Matthias

    Interesting application!

    The steps you mentioned about adjusting the contact radius after the meta-particle is formed is what I would have expected to resolve your problem.  

    Are you willing to share your simulation deck so that I can have a look?

  • Matthias Vink
    Matthias Vink Altair Community Member
    edited March 26

    Here is a zip folder containing all the files

  • Matthias Vink
    Matthias Vink Altair Community Member
    edited March 27

    I am also using the timoshenko beam bond model to see if that can handle large bond strain. But I am running into the following error:
    image

    I got the model from:
    https://community.altair.com/community/en/edem-api-contact-model-example-timoshenko-beam-bonded-particle-model?id=kb_article_view&sysparm_article=KB0121950&sys_kb_id=f73ef0c4dba6e910cfd5f6a4e296197a&spa=1 

  • ian shen
    ian shen New Altair Community Member
    edited August 5 
    I now use bonding v2 model to generate bonds, but the following problems occur how to solve this problem thankss

    image

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited August 8
    Matthias Vink said:

    Here is a zip folder containing all the files

    Hello Matthias,

    I am having a look at you simulation deck.  Please give me a few days to review it and get back to you.
    Kind Regards,

    Eric

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited August 9
    Matthias Vink said:

    Here is a zip folder containing all the files

    Hello Matthias,

    I had a look at your simulation deck.  I would suggest using a contact radius of 40% larger than the physical radius to ensure that the bonds are formed between the spheres on the outer sides of the chicken breast.  I also found that I needed to increase the stiffness of the bonds by about an order of magnitude to ensure that the bonds were stiff enough that the spheres would not separate beyond the contact radius in the simulation you had provided. 

    I suggest that you likely need to calibrate the bonds based on physical tests of something like a cantilever test or a flop test in which the angular deflection of the chicken breast is measured and used as the target to calibrate the normal and shear stiffness.

    You are able to change the contact radius of the particles after they are formed, however if the stiffness is too low the particles will not be held in place and bond failure will be unavoidable.  Note that increasing the contact radius does come at the cost of more contacts being calculated even though physically there particles as not in contact.  Typically you want to keep the contact radius as small as possible while also being able to replicate the behavior of the chicken breast.

    I have attached your simulation deck with the modified contact radius and normal stiffness.

    Kind Regards,

    Eric

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited August 9
    Matthias Vink said:

    I am also using the timoshenko beam bond model to see if that can handle large bond strain. But I am running into the following error:
    image

    I got the model from:
    https://community.altair.com/community/en/edem-api-contact-model-example-timoshenko-beam-bonded-particle-model?id=kb_article_view&sysparm_article=KB0121950&sys_kb_id=f73ef0c4dba6e910cfd5f6a4e296197a&spa=1 

    Matthias,

    I do not receive the error you are receiving when I use the Timoshenko model, which I obtained from the same location you listed above.  Perhaps you downloaded a older version of the model?
    Regardless, I would not recommend using this model for chicken fillets.  I think the Bonding V2 would be better. 

    We are working on a few new bond models that would be even better for this application.  I will see if we can share a beta version of the model with you.
    Kind Regards,

    Eric

  • Eric Veikle_22252
    Eric Veikle_22252
    Altair Employee
    edited August 9
    ian shen said:

    I now use bonding v2 model to generate bonds, but the following problems occur how to solve this problem thankss

    image

    HI Ian,

    I suspect that you have not enabled the contact radius for your particles as shown in the attached EDEM GUI.
    Kind Regards,

    Eric

Categories