Learning Forum

[darshankumar.bhat]

                                                 

[peteroznewman]

Look at the location of the element that is not converging.  Is it near a Fixed Boundary?  Changing the boundary condition or geometry may help convergence.

If you get 3 bisections in a row on the first substep, that suggests the applied load in step 1 is too large. Try breaking the model into a 2 step analysis.  In Step 1, make the load 1% of the final load.  Change the Initial and Maximum Timestep to 0.001 s and the Minimum Timestep to 0.00001 s.  These values may be extreme, but you must get the first substep to converge. If it easily converges, you can back off from these extreme values.

I assume you have Large Deflection turned on under the Analysis Settings.

[darshankumar.bhat]

Hello,
thanks for the reply.
I am already doing as you mentioned (dividing into mutliple load steps)
Here is my commands,
! Step 1: Just for the contact definition. I go upto 0.1% of the maximum load
    first_rampup_Time = 10 !seconds
    RATE,off !switch implicit creep off to get the contact
    kbc,0 
    nsubst,10,300,20,off
    cmsel,s,FORCE_NODES $ f,all,fy,NodalForce*0.0001
    time,first_rampup_Time*0.0001  $ alls  $ solve 
! Step 2: 1st force-ramp up.
    nsubst,1000,100000,1000,on
    kbc,1 !Loads are step changed (stepped): Useful for rate-dependent behavior
    RATE,on ! switch implicit creep ON
    PRED,OFF
    cmsel,s,FORCE_NODES $ f,all,fy,NodalForce*0.0002
    time,first_rampup_Time*0.0002  $ alls $ solve

With this config, I am experience this material convergence error along with L2-norm error.