Learning Forum

[RoyalFlowers]

Hello every one,

I have a question about contact detection methods.

Does the Gauss point which is a default contact detection method work for all situations?

Recently I have faced a model with surface-corner contact (as following pictures) and realized that Nodal detection method (Normal to Target) provide different results than Nodal-Normal to Contact and Gauss Point methods.

 

It is clear that results should be different between contact normal and target normal because smoother surface should be selected  to define normal directions. And in this case our contact area is a facet surface which results in multiply normal directions and software considers the average of them. But I cannot understand why results of Gauss Point method are different. And could we use Gauss Point as contact detection method for this case?

I would be thankful if someone could advise me in this regards.

Best regards, Shabnam Samsami

[Sandeep Medikonda]

See, section 3.9.6 in the help. Depending on other settings such as normal stiffness, frequency of update, trim contact etc the restoring forces can be different which can explain the differences. 

[RoyalFlowers]

Dear Medikonda,

Thanks a million for you answer regarding this point. The all settings of model were the same and I only changed the contact detection method from Nodal-Target to Gauss Point. And the displacement of loading point was changed from 1,38 mm  for Nodal-Target to 1,18 mm for Gauss Point. And the corresponding experimental value is 1,5 mm. Now I do not know why two different contact detection method resulted different outputs and which one should be taken. Also, could you please explain me what is restoring force?

Moreover, As I read  "for a model with an edge or corner in contact we may need to consider contact detection at the node locations. In Figure 1 the red circles represent pin ball radius around the detection point. With integration point detection option, the element corner penetrates into the target element before contact is detected". Hence how can I check the penetration before contact detection (as figure 1) to confirm that Gauss detection method can not identify the edge of contact surface?

 

 

I am grateful of your advice in advance.

Sincerly, Shabnam 

[Sandeep Medikonda]

Think of restoring force as a spring, which is pushing back against penetration and different detection methods can report different values. There are a lot of details that can contribute to the differences. Its hard to pinpoint one. For example a normal at the node point might have a slight different direction from the one at the gauss point. Hence the penetrations can be different which can cause a different stiffness to be applied when pushing back.

You can use the contact tool and contact results to understand how the Gap, stiffness etc are changing for a particular contact. This might give you more insight. 

[RoyalFlowers]

Dear Medikonda,

 

Thank you for the reply.

 

My concern was raised when I found that Gauss point  results was almost 15% different than nodal target method. Then I found that for surface-corner contacts nodal contact detection methods are recommended because Gauss point detect integration point which is located inside of element and cannot detect edge or corners which can be detected by nodal detection method. But I wanted to check that Gauss point cannot work for Surface -corner contact. However, still I am not completely sure about that. Is that true that we should not select Gauss point method for surface-corner contact? I really grateful if you can advise me in this regard, to be sure that my interpretation about these two different methods are correct.

Unfortunately because of website error I cannot insert the pictures and put them in the attachment. 

 

In picture 1, load and and boundary conditions with detailed view on contact site which is surface to edge is shown. 250 N axial force is applied on the remote point that has free y rotation and z displacement. Picture 1 also shows the details of contact in which Asymmetry friction-less contact with normal stiffness=0,1 and update in each iteration; adjust to touch; is applied and detection method was Gauss or Nodal Normal Target.

Picture 3 shows the initial contact table. As you see in Gauss point there is no initial penetration although with normal stiffness 0,1 we should have some penetration. Hence, I think that Gauss point cannot detect the edge or corner of contact surface because springs of contact area were not activated and we had no penetration.  

In Picture 4 contact reaction force is compared between two detection methods. Because normal of contact surface which is calculated in Gauss point (is that correct?) is different than normal of target surface, two different method resulted in different contact reaction forces. Contact Reaction force of Gauss pint was (-36 N, 10 N, 250 N) , although it was  (-41 N, 58 N, 250 N) for Nodal Normal target. The external load was 250N, and I do not know whether it is acceptable that resultant contact force for nodal normal target method is around 260 N .

Higher reaction force in y direction for nodal normal target method indicates that corner of contact surface is detected and object was pushed a way in y direction. According to Picture 5, there is higher penetration for nodal normal target method because corner of contact surface is detected and contact spring is activated and because of low normal stiffness value higher penetration is allowed. Since the contact springs are detected in nodal normal target method and displaced, the main object showed a lower deformation in X,Y and Z directions (which are shown in picture 6- comparing with Gauss method. But maximum Y deformation which occurred at the bottom of object is higher in nodal normal target method than Gauss method, because nodal normal target method resulted in higher contact reaction force in y direction. Moreover, because contact springs were deformed more in nodal normal target method the displacement of loading center was higher than Gauss point. in Gauss method displacement of loading point was (0 mm, 0 mm, 0.43 mm), although Nodal Normal target Method showed (0 mm, 0 mm, 0.45 mm).

I appreciate your comments in this regard in advance.

 

Sincerely, Shabnam

 

 

 

 

[RoyalFlowers]

Dear Medikonda,

Because of website error I cannot add my pictures to the discussion, could you please advise me about that.

Sincerely, Shaman