Crushable Foam

[hrtuwair]

Hi all,

I am a new user of ABAQUS and am trying to model a cube of sandwich panel, as attached. The stress-strain of the foam was obtained from a compressive test. Everytime I run the analysis, it takes too much time for the analysis and then got this error. " TIME INCREMENT REQUIRED IS LESS THAN THE MINIMUM SPECIFIED".

Thanks in advance for your help.

Hesham

 

[acspain]

Hesham,

Try adding 0 in your 'Displacement' boundary condition for degree of freedom 1 and 3. The top piece may be unstable and quickly deforming in those two directions.

Use C3D8R elements to begin with, C3D8I elements are computationally expensive. You can do a result comparison later once you get the model up and running.

The initial time step shouldn't be any smaller than 0.001. Increment sizes less than 1e-5 means there is something wrong with the model and solution time will be very, very long if it ever solves.

Review your *.dat and *.msg files for warning and error messages. These will give you clues as to what the problem may be.

You would also have better luck getting answers by posting this in the Abaqus forum.

Good luck.

 

[hrtuwair]

Thanks a lot for those tips. I will try them and update you. Another question please, regarding the stress-strain curve I got force-displacement curve from flat-wise compression test and convert it to stress-strain by dividing the applied force by the specimen area and deflection by the thickness of specimen. Is this way acceptable? or should I measure the strain in middle region of the thickness.

Regards

Hesham

 

[acspain]

Hesham,

It has been a long time since I modeled foam and at a former employer.

That said, I think your data manipulation is okay so long as you remember that once the material goes solid the test data may be questionable. Assumptions of uniaxial compression are being compromised. Additionally if test forces are high, test machine compliance will need to be accounted for (if accuracy becomes an issue between product analysis and bench test results).

I noticed in your model there were some material calibrations. I haven't used that functionality so I am not sure if it covers my next suggestion.

To check your material model, create a single element model with the foam material model and compress the element. Then compare your force displacement data between the model and test. Be careful of your boundary conditions on the 8 nodes and how you collect your force displacement data. Reference 'Abaqus Benchmarks Manual', Simple tests on a crushable foam specimen; Abaqus 6.12 section 3.2.7. May be a different section number in the release you are using but should have the same title.

Good luck.

 

[hrtuwair]

Thanks for your recommendations. Most of your suggestion helped me a lot especially changing the element type. I attached the file again as I did some modifications. Interestingly, when I put the poison ratio zero, the force-displacement curve I got from ABAQUS matched the experimental curve and it ran in a quicker time. However, when I change the poison ratio to be 0.35 (which is the right value), the analysis runs for hours and stops. I got this message (too many attempt was made) I tried to make the initial and minimum step too small but did not work. Any advice please.

Another question please, I do have also the tensile curve from experimental for this foam and don’t know where to put it in the model. Thanks
Regards

 

[acspain]

Hesham,

A Poisson's value of 0.35 for a foam is too large. Remember that the Poisson's value is defined as the negative ratio of the transverse strain over the axial strain. When a foam is crushed there is very little to no strain in the transverse direction, the material collapses straight down with little to no bulging of the sides, therefore Poisson's value is 0.0 or close to it. Now 0.35 may be a valid value for a solid form of your material but not the foam form. Since you now have a simulation material curve that matches your experimental curve then your material model has been validated for that mode of deformation.

At this point you are going beyond my experience. However a quick look at your model and the manuals, indicates that tensile data may not be acceptable in this model as it is intended for plastic compressive deformation. A material model that will accept tensile data is the Hyperfoam model, this model is used primarily with polymer foams that will rebound back to its original shape when released.

I hope this is helpful.
Have a good day!

 

[Demon3]

I am amazed at how much time people waste fitting a model to the data. A model is, in my opinion, only of use if it predicts what will happen so you don't have to do the experiment. So many books are filled with useless equations which just confuse people. The data is the real world.

Professor Jim Feast once had some wise words on modeling but I don't want to get into trouble quoting him.

Chris DeArmitt

Expert consulting & training

http://www.phantomplastics.com

 

[IRstuff]

I'll respectfully disagree on that. While data is the real world, there are plenty of systems that cannot be fully tested or observed. In those cases, which I think are quite numerous, we resort to the most accurate model possible, to ensure that our predictions are at least semi-plausible. In EE land, SPICE is used as a simulation (prediction) tool for electronic circuits. It would be impossibly difficult to perform actual measurements on most state of the art integrated circuits, since even the mere presence of a test probe would drastically corrupt any measurements made. That said, modeling is only part science; the other part is intuition and feel, that determines how many degrees of freedom to allow, to minimize the noise, while maintaining reasonable verisimilitude.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Demon3]

You mention a case where modeling allows you to do something and avoid the testing, which is the case I mentioned as making sense. So we agree.

Most modeling work is a pointless attempt to model something which brings no new understanding.

Chris DeArmitt

Expert consulting & training

http://www.phantomplastics.com

 

[hrtuwair]

Thanks, you’re right. The Poisson ratio after the elastic collapse stress is reached and exceeded, i.e. on the plateau part is acceptable to use the value of zero (and I already did that in my model as attached in my previous post), but prior to crushing, in the linear stress-stain range, the Poisson ratio for a closed cell foam is recommended to be taken equal to 1/3, according to the classical book by Gibson and Ashby.

Therefore, I don’t have any explanation why the analysis is terminated when I use a value of Poisson ratio? Any help will be appreciated.

Hesham

 

[acspain]

Hesham,

Interesting (read, I don't know why), I personally would not worry about the Poisson ratio so long as there is a good match between the material model and experimental data as you indicated that you acheived. Poisson's ratio, in my experience, plays a role only when the part is volumetrically constrained. For example an o-ring in a gland that is too tight. A little heat and wala the shaft has seized because the o-ring has expanded to fill every crevice. Force output from that analysis would be sensitive to the Poisson's value (along with other effects), and difficult to achieve experimentally. Now if the same o-ring is in a bigger gland (not volumetrically constrained), you could change the Poisson's value and the force output would not be affected as the part is being deformed/deflected not compressed from all sides.

Demon3,

I am assuming the Hesham is working towards modeling a complex structure and/or complex loading. I am employed as a Finite Element Analyst because the time spent modeling materials and parts saves thousands of dollars and huge amounts of time by reducing the number of prototypes built and answering questions that can not be answered through experimentation. Additionally, research and developement into new modeling techniques may or may not be adopted into widespread use but some of the models currently employed, especially within elastomers, have significant shortcomings when utilized in product engineering. Regardless, so long as the shortcomings are understood, the models still produce answers and reduce the number of prototypes needed to bring a product into production.

IRstuff,

Thank you for your well worded support.

Have a good day!

 

[hrtuwair]

Thanks again, Acspain,
I see what you mean. At this stage the Poisson ratio might not be an important key but am sure it will be for my later models. I believe it will play an important role as the foam I am verifying now will be the core a of sandwich beam structure, which the shear rigidity of the foam will be estimated based on the Poisson ratio, that is why I am trying to include it now to make sure everything works fine at this stage and then will move forward to my next stage. Thanks again for your effort.
Hesham