"tangent" stiffness matrix 1

[seihin]

Hello friends,

I am running a static, nonlinear geometry analysis of a 2D grid structure. The grid (kagome grid, made up of equilateral triangles) has about 36000 members each with five nodes, making the analysis very big. What I am trying to find out is the behaviour of the structure when I actuate one member in the middle. The aim is to plot force-displacment graphs to investigate the change in stiffness of the structure due to nonlinear effects. We hope to load the structure until it fails by heating the centre bar. However the analysis always fails at about 40% actuation strain, with the following error messages.

***WARNING: THE SOLUTION APPEARS TO BE DIVERGING.

***ERROR: TIME INCREMENT REQUIRED IS LESS THAN MINIMUM SPECIFIED. ANALYSIS ENDS

The same error pops up even if i set the increment to impractically small values, say 1e-9. Does anyone have an idea of what seems to be the problem?

I tried to investigate this problem by looking at the stiffness of the structure at each step of the analysis (ie. the stiffness at lower actuation strains). I wish to extract the samllest eigenvalue of the "tangent" stiffness matrix at each step. However, ABAQUS doesn't seem to have an option to get stiffness matrix eigenvalues. Does anyone know how to do that? (I can get the "tangent" stiffness matrix itself by using the *ELEMENT MATRIX OUTPUT option, but since my model is so huge it is not so practical as the resulting fil file can be up to 2GB, it takes ages just to load that into Matlab. All I need is the smallest eigenvalue of that matrix).

Many thanks!

 

[amaratre]

Hello,
I am doing a contact analysis and I am encountering the same problem. First of all did you look in your message file for what seems to be happening? I tried to start my analysis with smaller time steps and also tried like you did by taking the min. time step to 1e-9, but that doesent work. From the message file it seems like abaqus tries to balance the force residuals but cannot do so, so its taking smaller and smaller time increments until it reaches the min. time increment where it aborts.
I am trying to see how I can resolve this. Maybe you can try and change the mesh? My analysis works for certain meshes.
Amar

 

[Drej]

My guess is that your "tangent stiffness matrix" will tell you nothing about why the error occurs. The "problem" is probably that you're trying to load your structure beyond a practical, physical limit or that your material properties, boundary conditions &c are incorrect. Also check you're time increments are sufficiently small if it's implicit and that autostepping isn't causing you problems because of some initial step size. Check these first, along with your current DOF displacements etc. (reality check) as well as an analytical solution.

Cheers,

-- drej --

 

[mike21]

Mabey you can try another software: "FEAPV",it can output the eigenvalue of the tangent matrix.