Help needed in Abaqus FEA

[victorroda]

Hi,

I'm doing a static analysis of a beam with a point load. I think it is a very simple problem, but Abaqus can't solve it. After five incrementes and a time incremente reduced to 3.906E-03, the *.msg file says:



***NOTE: THE SOLUTION APPEARS TO BE DIVERGING. CONVERGENCE IS JUDGED UNLIKELY.
***ERROR: TOO MANY ATTEMPTS MADE FOR THIS INCREMENT

Can anyone help me with this? I attached the input file if it can helps to find the problem

Thanx in advice

 

[mrgoldthorpe]

Replace the relevant commands by the following. This simply restricts the increment size to 2% of the total load.

*STEP,UNSYMM=YES,NAME=STEP1,NLGEOM=YES,INC=2000
*STATIC
** init'l time period min. max.
** time of step time time
0.02, 1.00 , 0.0000001 , 0.02

 

[InvariantL]

I don't think reducing the increment size is going to help here - the solver has already automatically reduced it to 3.906e-3 and it still doesn't converge.

There are a thousand reasons this can happen, but you might not have adequately constrained the structure, i.e. it might be free to spin about the 3-axis, depending on where you applied your bc's.

Try replacing the load with a prescribed displacement in some direction, just for debugging. If the problem is with the bc, this will let the analysis complete, and then you can look at the results to see if you have constrained it as intended.

http://www.invariantlabs.com

 

[corus]

It says you have one negative eigenvalue, which means it's not restrained in one direction. Looking at the model I think it can spin on uts axis. To stop this you could model half the axle and use symmetry to prevent spinning.

Tata but not yet tara

 

[mrgoldthorpe]

InvariantL, I ran it as I instructed (V692) and it completed.

Often, when increment sizes reduce too much, the displacement corrections are of the same order of size as displacements themselves, which prevents convergence. My suggested fix was to prevent ABAQUS taking large increments and so avoiding cut-backs.

Agreed Corus, there could be inadequate constraints, but I didn't even look at the model in CAE.

 

[johnhors]

"there could be inadequate constraints" -

There are inadequate supports!

Only five degrees of freedom have been restrained in the inp file.

For a 3D model a minimum of six degrees of freedom (in a 3-2-1 arrangement) are required to be restrained to remove all of the rigid body motions.

http://www.Roshaz.com

 

[InvariantL]

"Often, when increment sizes reduce too much, the displacement corrections are of the same order of size as displacements themselves, which prevents convergence."

Yeah! I forgot about that though I've observed it many times and never quite understood it. It's surprising that you hit the accuracy floor of the corrections at increment sizes that are so commonly reached in nonlinear analyses. But there you go.

I assume your run completed even without changing any of the constraints.

http://www.invariantlabs.com

 

[mrgoldthorpe]

InvariantL, yes it completed without further modifications. It cut-back a little towards the end.

The OP seems to have lost interest!

 

[victorroda]

Hi all, and thanx for the replies

The problem was i haven't constrained the 6th degree of freedom.

Sometimes we look for the answer so deep, when it really is very simple.

Thank you very much

Víctor