contact problem

[masood4626]

I want to model a steel silo under earthquake excitation.
I used a umat subroutine for constitutive behavior of granular material (hypoplastic mateial with inergranular strain) I tried to model a steel silo with granular material under earthquake excitation with Abaqus standard. I modeled silo walls and bottom with 4 node shell elements and granular material with 8 node solid elements then I used surface to surface contact between silo walls, silo bottom and granular material, when I use default hard contact for normal behavior of contact in Abaqus standard I have overconstraint problem due to using contact pair and earthquake motion applied to the silo bottom as acceleration boundary condition together, and when I use penalty or augmented lagrange method for normal behavior of contact, I encounter two type of warnings

in first step that I apply gravity load the waring is

1)During back substitution 625 elements encounter a load applied to a degree of freedom that does not have any stiffness. The elements have been identified in element set WarnElemBackSubNoStiffStep1Inc1.

and in second srep the warning is

2)There are 2 unconnected regions in the model.

when I increase stiffness scale factor the first waring is ommied but in the second step I have warning that says

There are 2 unconnected regions in the model.

in fact solid elements penetrate into silo wall and my model becomes unstable. please help me I am completely confused.

 

[rstupplebeen]

Try starting with a simple contact analysis with only 2 parts and without a umat. 2 steel blocks is where I would start. I hope this helps.

Rob Stupplebeen

 

[masood4626]

now my model have only two parts. I have used drucker prager model instead of umat before for modeling of granular material and these warnings still exist.

 

[corus]

Look in Viewer at your results to see where the elements with no stiffness are. It sounds like a numeric problem though if the warning disappears with a change in material.

If one body passes through another then you haven't made contact for whatever reason. As in all these cases, start with a dummy load case of fixed displacements to ensure some contact is made before applying your real loads.

corus