Learning Forum

[ElCid01]

Hello everybody

I have a problem concerning the definition of a paired-based frictional contact between a plate (shell181 with conta173) and a fixture (solid185 with targe170). In the code below you can see how I set the contact up. My problem is that regardless of whether I define a linear (bonded) contact or a frictional contact, if I check the contact status with cncheck, the message says always that the contact is not closed. There is also no geometrical penetration or gap which could cause the contact to be open. This happens only with the Augmented Lagrange or Penalty method setup, if I change it to MPC-contact (keyopt,6,2,2) then the contact closes. The problem is that with MPC I cannot define a frictional contact.

Did anybody have a similar problem and can give me some advice?

For any suggestions I would be very happy.

Best regards ElCid

! Contact element
ET,6,CONTA173

TB,FRIC,1,,,ISO
tbtemp,0
TBDATA,1,0.1

TB,FRIC,4,,,ISO
tbtemp,0
TBDATA,1,0.1

! key-options for contact elements (ID = 6) -> keyopt sets element key options
keyopt,6,1,0
keyopt,6,2,0    !Augmented
keyopt,6,12,0  ! Element type ID (4), No. of keyoption to be defined (12 = bonded), value of keyoption
keyopt,6,5,3    ! Auto gap,penetration correction

r,1500
real,1500

allsel,all,all

csys,0

esel,s,type,,5
nsle,s
nsel,r,loc,z,0
mat,4
Type,3  !Targe170 element defined previously
ESURF

r,1500
real,1500

csys,0
esel,s,type,,1
esel,r,cent,z,0
nsle,s
csys,1000
nsel,u,loc,x,0,bc_rad_free
csys,0
cmsel,u,bc_nodes_bot,node
Mat,1
Type,6
Esurf
csys,0

allsel,all,all

[John Doyle]

Try adding:

KEYOPT,6,11,1   

This will include shell thickness effect which is not needed with bonded MPC

Regards,

John

   

[ElCid01]

Hello John

Thank you for your answer.

Unfortunately it still doesn't work.

I have attached a picture where the elements and the message is visible.

 

[John Doyle]

Since no contact is detected, can you check your element normal directions?  

The contact and target normals need to face each other for nonlinear contact to engage.

You can reverse the normals with 'ESURF,,reverse'.

If this turns out to be the issue, you can correct your original script with 'ESURF,,bottom' for the contact creation, so the normal directions are correct at the time of creation the next time you read in your input.

Since bonded works ok, your pinball is sufficient.

Regards,

John

 

[ElCid01]

That I have already tried. It doesn't show any improvement. I really don't know what else it could be. It is really strange. The pinball radius should be ok because it is identical with MPC or augmented lagrange. Only changing keyopt,6,2,0 to keyopt,6,2,2 the contact closes...

[John Doyle]

Also, try a displacement based load for stability and make sure you are using a small enough time step so as not to step right over the initial gap in first substep.

Also, if this is a force (pressure) based load, you might need to close the initial gap also with KEYOPT,,5,1.  Keep KEYOPT,,11,1 to account for shell thickness as well.

[ElCid01]

KEYOPT,,5,3 is already defined. What I noted is that with MPC contact a TOLS parameter is defined, which in augmented lagrange contact is not defined. Could that have an influence?

 

[ElCid01]

Hello John

You were right: I changed the direction of the normals of the contact elements with esurf,,bottom and then the contact closed.

Thank you very much for your help.

Best regards