Defining contacts

Bocaj22

Hello,

What is the current best practice for defining contacts for Radioss in Hyperworks? Our models will generally have Type 2 and Type 7 contacts. I usually use sets (grnode and surfseg or surf_ext) to identify the regions where the contact will occur. However, in examples in the documentation, they sometimes identify only the component id's instead. This seems a lot faster, but are there are any downsides?

Thanks.

Mathis

Hi Bocaj22, 

I don't think there is any downside to this practice. Whenever you can select a full part for a contact, it is better to use a /SURF/PART/EXT or a /GRNOD/PART. It is faster to setup and it also allows you to remesh your part with HyperMesh without having to redefine your surface or grnod set. 

Btw, did you try to use /INTER/TYPE25 instead of /INTER/TYPE7, it can be an interesting alternative (bigger time step in general) and possibility to handle solid2solid contacts.

Regards,

Mathis. 

Bocaj22

Mathis Loverini_20748 said:

Hi Bocaj22, 

I don't think there is any downside to this practice. Whenever you can select a full part for a contact, it is better to use a /SURF/PART/EXT or a /GRNOD/PART. It is faster to setup and it also allows you to remesh your part with HyperMesh without having to redefine your surface or grnod set. 

Btw, did you try to use /INTER/TYPE25 instead of /INTER/TYPE7, it can be an interesting alternative (bigger time step in general) and possibility to handle solid2solid contacts.

Regards,

Mathis. 

Hi Mathis,

Thank you for your response. Is the ability to select a full part for contact relatively new? I wonder why many older tutorials involve selecting individual nodes/elements rather than the full part. Is it more efficient to run the simulation if you limit the parts of the model that are considered for contact?

I haven't used type25, but I will look into it more.

Thanks,

Bocaj22

Mathis

bocaj22 said:

Hi Mathis,

Thank you for your response. Is the ability to select a full part for contact relatively new? I wonder why many older tutorials involve selecting individual nodes/elements rather than the full part. Is it more efficient to run the simulation if you limit the parts of the model that are considered for contact?

I haven't used type25, but I will look into it more.

Thanks,

Bocaj22

Hi Bocaj22,

I have always known this feature to be honnest. I guess the tutorial you are refering are eather outdated or just aim to display different ways to setup contact. Are these tutorials from the Altair documentation? 

 "Is it more efficient to run the simulation if you limit the parts of the model that are considered for contact?"
==> I guess it is a bit more efficient, but it will depend on case to case. 

==> A global self-contact (with option Igap=3 for automatic gap value) should work well and should be efficient in term of setup. 

I hope it helps.

Regards,

Mathis. 

Bocaj22

Mathis Loverini_20748 said:

Hi Bocaj22,

I have always known this feature to be honnest. I guess the tutorial you are refering are eather outdated or just aim to display different ways to setup contact. Are these tutorials from the Altair documentation? 

 "Is it more efficient to run the simulation if you limit the parts of the model that are considered for contact?"
==> I guess it is a bit more efficient, but it will depend on case to case. 

==> A global self-contact (with option Igap=3 for automatic gap value) should work well and should be efficient in term of setup. 

I hope it helps.

Regards,

Mathis. 

Here is a recent post from Altair, for example.

https://www.youtube.com/watch?v=r4ZnE1ouoiQ

In this scenario, it would be simpler to use /GRNOD/PART and /SURF/PART/EXT right?

There are many examples in the documentation. Looking closer, there does seem to be a variety of ways to setup contacts.

Mathis

I think in that Altair how-to case, they want to tied a specific surface to a specific set of nodes. And I think it is the best way that is shown. 

But, as you were orignally talking about "whenever you want to select a full part for a contact", I assumed you want a contact between two full parts. 

As an example it is shown here how to do a full self contact :https://www.youtube.com/watch?v=kQnMhFM9iAE.

Regards,

Mathis

Bocaj22

Mathis Loverini_20748 said:

I think in that Altair how-to case, they want to tied a specific surface to a specific set of nodes. And I think it is the best way that is shown. 

But, as you were orignally talking about "whenever you want to select a full part for a contact", I assumed you want a contact between two full parts. 

As an example it is shown here how to do a full self contact :https://www.youtube.com/watch?v=kQnMhFM9iAE.

Regards,

Mathis

Thanks for sharing the link. For a type 2 contact in particular, we certainly know which surfaces and nodes should be attached. But like you said before, defining the contact via the parts makes setup much easier, especially for situations where things are automated and/or we may want to remesh. However, we need to confirm that results are the same between the two methods.

Mathis

bocaj22 said:

Thanks for sharing the link. For a type 2 contact in particular, we certainly know which surfaces and nodes should be attached. But like you said before, defining the contact via the parts makes setup much easier, especially for situations where things are automated and/or we may want to remesh. However, we need to confirm that results are the same between the two methods.

Hi Bocaj22,

If you want to tied all the nodes of a part to the surface of an other part, then it is the same. but be mindful, that for type2 interface, you need to have projection of secondary node to main surface. If it is not the case, you should "play" with the ignore flag of the interface to remove automatically the nodes that don't project to the main surface. 

Regards,

Mathis