ABAQUS/Explicit - Excess Kinematic Energy

[pirs]

Hello!

I am trying to model a deformable steel beam which contacts with two rigid plates: one acting as a support and the other acting as a loading plate through applied displacement. I am using the *contact algorythm to perform a quasi-static analysis.

However, when observing the results, several sudden increases on the support reaction occur followed by quasi-horizontal platforms, either when applying the displacement with a linear increment or smooth step.

Analysis of the kinematic energy vs. total energy shows that there is quite some influence of the kinematic term, even though I am trying to perform a quasi-static analysis.

My question is: how does one avoid such sudden reaction increases, which appear to result from excess kinematic energy?

I have tried both using *bulk viscosity and *fixed mass scaling, but nothing seems to help, although the results might alter a little.

Attached is a graphic of the time/support reaction.

Thanks in advance to any input on this.

 

[Dave442]

Hi pirs,

What kind of loading rate are you using for the analysis. For a quasi-static analysis the kinetic energy should remain below 5% of internal energy throughout the majority of the analysis.

If this is not the case I would remove any mass-scaling and reduce the loading rate by increasing the solution time. Also, the smooth step amplitude should help reduce any fluctuations.

There is a good tutorial on quasi-static procedures in Section 13.5 of the "Getting Started with Abaqus: Keywords Edition" tutorials manual that comes with the software.

Good luck,
Dave

 

[pirs]

Although I am able to keep the kinematic energy below 5%, I observe that it peaks considerably during the initial stages ( of the test, reaching up to 15-20% of total energy. This is what I have found to be unacceptable, as you pointed out.

For the amplitude function I have defined as follows:

*AMPLITUDE, NAME=Ramp
0., 0., 100, 1,

Given this, the dynamic step is also 100s long:

*DYNAMIC, EXPLICIT
, 100
*BULK VISCOSITY
0.06, 1.2

As for the mass scaling, I have used the following parameter:
*FIXED MASS SCALING, TYPE=UNIFORM, DT=0.001

Any further input you can give me on these values? I will read the reference you gave me on quasi-static procedures and perhaps reach for a conclusion myself.

Thanks for your help!

 

[Dave442]

Hi pirs,

Mass scaling will increase the amount of kinetic energy observed during the analysis so I would suggest running the analysis with no mass scaling first.

I'm not quite sure I understand your analysis. It sounds like you are performing a compression type analysis. Do you really need to model the support and load plates? Could you constrain one end of the beam and apply displacements directly to the other end?

Dave

 

[pirs]

Well, it is indeed a compression analysis.

I don't necessary need to model the support and load plates, but to some extent the contact interaction between the load application plate and the support plate with the beam are relevant, thus I decided to include them in the model.

I have tried many different options and I always obtain the sudden reaction increases, even with low kinematic energy. I am thinking that one other reason for this might be due to the contact itself... but that is a whole different question.

Anyway, thank you for your help.

Pirs

 

[pirs]

Just to add something else to my previous post:

Modelling the rigid plates is important since I want to observe the behaviour of the beam's web-flange corner near both rigid plates.
If I apply a simple displacement to the top flange and constrain the bottom flange I will most probably have to neglect modelling the behaviour of said corners, in terms of the interaction with the rigid plates.

Pirs