Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Contact pairs passing through each other
Posted 17 ago 2010, 14:21 GMT-4 10 Replies
Please login with a confirmed email address before reporting spam
Hi,
I'm in the process of trying to create a COMSOL model involving contact between two surfaces initially separated by a non-solid domain.
I've actually got two models. In one model, I was able to make this contact pair work, so I know that it is possible. In the other model, COMSOL allows the surfaces to pass through each other without resistance. I have tried several times starting over, deleting physics modes and rebuilding, adding contact offsets and changing penalty factors. None of these seem to have any effect. the contact pair boundaries simply pass through each other.
I cannot figure out why contact in one model works and fails in the other. Both are created using the same physics modes, similar topology ect. Anyone know what I'm doing wrong in the one that doesn't work?
*Note: I'm aware that the contact pair in the 'working' model eventually leads to non-convergence. That is a separate issue that I think I can solve.
I'm in the process of trying to create a COMSOL model involving contact between two surfaces initially separated by a non-solid domain.
I've actually got two models. In one model, I was able to make this contact pair work, so I know that it is possible. In the other model, COMSOL allows the surfaces to pass through each other without resistance. I have tried several times starting over, deleting physics modes and rebuilding, adding contact offsets and changing penalty factors. None of these seem to have any effect. the contact pair boundaries simply pass through each other.
I cannot figure out why contact in one model works and fails in the other. Both are created using the same physics modes, similar topology ect. Anyone know what I'm doing wrong in the one that doesn't work?
*Note: I'm aware that the contact pair in the 'working' model eventually leads to non-convergence. That is a separate issue that I think I can solve.
10 Replies Last Post 30 ago 2010, 07:53 GMT-4
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I got a reply from COMSOL support about this issue, it is copied below:
"I realized that the solid mechanics physics was off for few domains.
Given that, the contact pairs were set correctly (It should be union).
However, in that model, both the solid domains are coming close to each
other and thus completely collapsing the domain in the middle. This is not
true in your other file, where only a part of the solid domains are coming
into contact with each other. You should remove the domain in the middle
between the solid objects (between the contact pairs) and not do heat
transfer calculations in that domain.
If you do want to include that domain, you can independently do heat
transfer calculation for just that one domain and couple it with the rest
of the model using coupling variables (extrusion coupling). This will be
more advanced but first you should just do the thermal stress and contact
analysis for just the solid domains."
Does this make sense to anyone? The domain in the non-working case does not collapse uniformly. It collapses faster in the middle, which to me resembles the working case. Can anyone explain the difference?
"I realized that the solid mechanics physics was off for few domains.
Given that, the contact pairs were set correctly (It should be union).
However, in that model, both the solid domains are coming close to each
other and thus completely collapsing the domain in the middle. This is not
true in your other file, where only a part of the solid domains are coming
into contact with each other. You should remove the domain in the middle
between the solid objects (between the contact pairs) and not do heat
transfer calculations in that domain.
If you do want to include that domain, you can independently do heat
transfer calculation for just that one domain and couple it with the rest
of the model using coupling variables (extrusion coupling). This will be
more advanced but first you should just do the thermal stress and contact
analysis for just the solid domains."
Does this make sense to anyone? The domain in the non-working case does not collapse uniformly. It collapses faster in the middle, which to me resembles the working case. Can anyone explain the difference?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I have made a small amount of progress. Based on the suggestion of COMSOL support above I have tried working without a meshed gap. I found that if the gap is meshed, the contact problem will not work. If the gap is not meshed, then I can make it work. This happens regardless of any equations being solved in the gap area.
Now I have to figure out why contact works with a meshed gap region in the working example but not in the failed example. Any thoughts?
Now I have to figure out why contact works with a meshed gap region in the working example but not in the failed example. Any thoughts?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
If what COMSOL support says is true, there must be some criteria which decides if contact pairs will be applied across a domain based on how 'collapsed' a domain is. Does anyone know what these might be or where I could read about them?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Anyone? Ivar?
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I FIXED IT! Actually, I found a bit of a work around, it isn't ideal, but it might help some other people out.
Apparently, COMSOL doesn't like contact between two parallel meshed surfaces. To get around this, I added a very small bump to one of the surfaces. Since the bump is so small (in my case 1e-7m) it changes the physics insignificantly but allows COMSOL to preform the calculation. I've attached the file so others can see.
NOTE: I'm not exactly sure why COMSOL behaves this way, I think that it might be to prevent the domain from collapsing entirely (which could result in problems like numerical instability and division by zero). Regardless, it seems to me like this behaviour was deliberately engineered into COMSOL, so be careful about circumventing it.
Apparently, COMSOL doesn't like contact between two parallel meshed surfaces. To get around this, I added a very small bump to one of the surfaces. Since the bump is so small (in my case 1e-7m) it changes the physics insignificantly but allows COMSOL to preform the calculation. I've attached the file so others can see.
NOTE: I'm not exactly sure why COMSOL behaves this way, I think that it might be to prevent the domain from collapsing entirely (which could result in problems like numerical instability and division by zero). Regardless, it seems to me like this behaviour was deliberately engineered into COMSOL, so be careful about circumventing it.
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I also have a problem with my contact pairs. There is also overlap between the two interfaces of the contact pair.
Anyone who knows what to do?
Anyone who knows what to do?
Attachments:
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Did you try what I described above? It might help if the two boundaries are parallel. If that isn't the problem than I you'll have to post a file for people to look at.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Did you try what I described above? It might help if the two boundaries are parallel. If that isn't the problem than I you'll have to post a file for people to look at.
The two boundaries are not parallel in my case. In my previous post there was already a file attached to look at.
Greetings,
Steven
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Sorry about that, I only read the email notification about the thread, not the actual post so I missed it.
Looking at your file, my only guess (and this is only a guess), is that COMSOL wont like the fact the boundaries start in contact. You might try modifying the polygone shape slightly to such that there is an additional small boundary that separates the big ones you want contact between. I don't know if this will work, but it might be worth a try. If you make the boundaries small enough, it shouldn't significantly affect the result, thought it will affect your mesh.
Let me know if this works out.
Looking at your file, my only guess (and this is only a guess), is that COMSOL wont like the fact the boundaries start in contact. You might try modifying the polygone shape slightly to such that there is an additional small boundary that separates the big ones you want contact between. I don't know if this will work, but it might be worth a try. If you make the boundaries small enough, it shouldn't significantly affect the result, thought it will affect your mesh.
Let me know if this works out.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi,
There is already a small gap between the two boundaries of my contact pair.
However, I did a new simulation in which I splitted the boundaries of the contact pair near the end and I chose the contact pair in that way that it was not extending all the way out. I thought that Comsol was having some troubles with the ends of the source and destination boundaries of the contact pair which were in contact. However, the two sides of the contact pair were still overlapping.
Steven
There is already a small gap between the two boundaries of my contact pair.
However, I did a new simulation in which I splitted the boundaries of the contact pair near the end and I chose the contact pair in that way that it was not extending all the way out. I thought that Comsol was having some troubles with the ends of the source and destination boundaries of the contact pair which were in contact. However, the two sides of the contact pair were still overlapping.
Steven