Surface contact forces, summed

[Anonymous]

Hello,

I'm using the method outlined here to get total contact forces for a bonded contact (surface to surface). Total forces summed for master nodes and slave nodes normally agree.

But apparently if some of the nodes on one of the selected surfaces  are shared between two contact definitions the results get crazy.

I assume that the contact force results for nodes that are used in more than one contact definition gets results that are somehow affected by by either or both contacts. Is that a correct assumption?

Could someone shed some light on how to interpret the results in this case?

Could it be that calculated stresses are also wrong if nodes are included in more than one contact definition?

I.e should I always make sure that no nodes are used in more than one contact definition?

I have had problems when nodes (not elements) are included in both a rigid body connector and a surface contact. 

The contacts are defined by BSCONP / BSSEG cards. The result I use is CONTACT FORCE (TX, TY and TZ).

I have a simple geometry demonstrating the problem I can share if anyone is interested.

[John_Holtz]

Hi @Anonymous 

It sounds like you are summing the contact forces for contact pair 1, summing the contact forces for contact pair 2, and then adding those two results together. Is that correct?

If that is what you are doing, the sum of the two is incorrect because you are summing the same nodes twice. Instead, you should get the node numbers for contact pair 1 and 2, remove the duplicate nodes per the article, and then get the sum of the contact force. This sum should be correct.

The stress results (and all the other results) are correct because there is only one contact force at the node. There is not one contact force at the node for contact pair 1 and a different contact force at contact pair 2.

A similar thing happens when looking at the SPC Summation in Inventor Nastran. The result shown is not the result at the selected constraint. The result shown is the reaction force in all directions at the nodes on the selected geometry. For example, this image shows 3 symmetry constraints on 3 faces. When looking at the SPC Summation for constraint 1 (Z symmetry), the nodes along edge A are also fixed in the X direction, so those nodes contribute an X reaction force. The nodes along edge B are also fixed in the Y direction, so those nodes contribute a Y reaction force. So although you have selected "constraint 1", the SPC summation has reactions in X, Y, and Z.

[Anonymous]

Hi John,

Thanks for your answer! I believe what I see is similar the SPC case you are describing.

But I think I was not clear about the forces I'm comparing. I'll try to elaborate.

With reference to your figure I have one Contact pair on face 1 (e.g "master" is face 1, "slave" is on a different part).

Then I have another Contact pair on face 2 (e.g "master" is face 2, "slave" is on a 3:rd part).

Then the nodes along edge A will be shared between Contact 1 Master and Contact 2 Master, right?

What I see is that if I have a contact pair with no shared nodes, the summed results for Master nodes and Slave nodes are virtually exactly the same (opposite signs though). This should of cause be the case; the total forces on one side of the contact must be the same as the sum on the other side.

But if there are shared nodes on one of the sides in a contact pair, the sum of forces is not "correct", i.e if I want to know how much force is transferred from Part 2 to Part 1 via Contact pair 1 I can't use the results obtained for Contact pair 1 Master, because the values for contact forces on nodes along edge A will be influenced by Contact pair 2 (net force sum for each node I assume??).

As long as at least one of the sides in a contact pair does not have any shared nodes I can use that side (master or slave) of the Contact pair to get the total force transferred through a contact.

Does this make sense??

If my understanding is correct, maybe this should be pointed out in the article describing the procedure?

/Olav

[John_Holtz · Accepted solution]

Hi Olav,

I think I understand your description. I think the problem is you need to use the contact force components and exclude directions that do not make sense. (Just like with the my constraint example. You cannot use the total SPC summation for the X symmetry constraint because the total reaction force on face 2 includes Z reaction force, and it is incorrect to use the Z reaction when trying to report the X reaction.)

If the above is an accurate understanding, then:

1. "Then the nodes along edge A will be shared between Contact 1 Master and Contact 2 Master, right?" Yes, that is correct.
2. "If I want to know how much force is transferred from Part 2 to Part 1 via Contact pair 1 I can't use the results obtained for Contact pair 1 Master, because the values for contact forces on nodes along edge A will be influenced by Contact pair 2"  You cannot use the total contact force on the face to get the correct result. You need to use the components and ignore some components. In the example of flat faces with no friction, it is easy to eliminate the components. If either of those is not true, it may not be so easy to eliminate the fraction of the contact face due to the adjacent face.. 

One solution would be to put a small chamfer on the edge. Then no nodes are shared between the contact pairs.

Hopefully, you can use the opposite contact face (on part 2 and on part 3) in your real example, assuming that they provide the contact force directions that you desire.

[Anonymous]

Hi John,

Yes, that's exactly my understanding. 

If at least one side of a contact does not share any edges with an other contact definitions I can use that side.

If both sides have edges that are shared with other contact definitions the geometry has to be changed in some way so there are no common edges on at least one of the sides.

/Olav

/Olav