Odd Force Output Implicit Analysis

Altair Forum User

I stumbled upon an unexpected behaviour regarding the force output in an implicit analysis.

I look at the following model of a honeycomb unit cell. On both ends there is a RBODY. While at the top there is an imposed velocity and the bottom is fixed in all 6 dof. The top RBODY is additionally constrained so that the downward movement is guided. (in case you are wondering, the nodes of the model are randomly distorted in order to achieve buckling also in implicit analysis)

I have been playing around with implicit and explicit solvers. Using the explicit solver I get very much an expected behavior of the force. I plotted both the SPC reaction force in Z (direction of IMPVEL) and the FZ of the RBODY at the top.
Both are more or less the same and it machtes my experiments.

However when I do implicit analysis i get odd results. First of all the FZ of the upper RBODY and the REACZ of the lower SPC are not the same. Secondly and most importantly, the force does not start at zero, which completely doesnt make sense.

When I do Non-Linear implicit I manage to get to initial force almost to zero, depending on what initial time step I set. The smaller the initial time step, the closer I get to ZeroForce at start.

However, the forces of the upper RBODY and the lower SPC are not the same. The force of the RBODY seems to match the explicit analysis. The SPC Force seems to be wrong.

Can anybody explain this to me? What do I do wrong? I have attached two picture (of the Model and of the plot) as well as the model files.

Regards
Ralf

Altair Forum User

This is the output I generated as well as a energy plot for the implicit analysis.

I am using Ishell=24.

I am not using a rigid wall because I wanted to have a supported compression.

regards
Ralf

Altair Forum User

Hi Ralf,

Perhaps we should use a smaller time step for the implicit for comparable results,

For the below implicit parameters,

we get a good correlation between the forces and the force increment starts with 0,

Please review and we can discuss further,

Altair Forum User

forces

can you please provide linear implicit parameters you use as in the first image of the above post?

Altair Forum User

thank you for your quick reply

I used slightly different engine file parameters for my implicit analysis (I am gonna try yours as well, when I have the time). But with my original settings, the compuational savings I have using implicit analysis vanish once I reduce the time step too much.

So I would like to keep a large time step so that the results are just sufficient.
And I am able to replicate the explicit results with a decent compuational time using the implicit solver.
However, the force doesnt start at zero, which in theory doesnt make sense to me.

I could live with the fact the force output depends on where I retrieve the data, as long as I know what is correct. Even though I would still like to understand why the force output at the lower SPC and upper RBODY are different. So an explanation would be nice.

Maybe you can also generally explain to me why the force output doesnt start at zero and why it depends on the time step.

My linear implicit settings are as follows:

/IMPL/QSTAT
/IMPL/LINEAR
/IMPL/SOLVER/1
1 0 3 0.0
/IMPL/PRINT/LINEAR/-1000

Altair Forum User

SeemannR

As you know, time step for an explicit analysis is calculated based on the Courant condition,
for implicit the time step is used rather to control the nonlinear iterations, i.e. for loading (or/and displacement) control;
Time step has no physical meaning for a static problem. There are many time step control parameters for implicit solution (/IMPL/DT, DT_ini, DT_min,DT_max) and they all might affect the results.
I will get back to you with more details soon,

Altair Forum User

Hi,

instead of the results from time history, /TH/NODE, REACX(YZ), request animation files results of Reaction forces for imposed velocities, displacements, accelerations and boundary conditions by using.
/ANIM/VECT/FREAC
/ANIM/VECT/MREAC

then you can load the h3d file in HyperGraph and plot the results. When using this method the forces will start at zero.

You could try using FOPT - Forces and moments for rigid bodies, rigid walls and sections, for your rigid bodies.
/ANIM/VECT/FOPT

For good results with implicit, it is necessary to have a large number (at least 50 but better to have 100 or more) of time increments. It is best if you can pick a time step that stays fairly constant.
As an example to get at least 100 time steps you could use these options,
/IMPL/DTINI
0.01
/IMPL/DT/STOP
# Dt_min Dt_max
1e-6 0.01