PDA

View Full Version : ABAQUS Tension Analysis



goldm066
2008-08-03, 18:29
Hello Everyone,

I am working on increasing the compliance of a hyper-elastic tube by cutting holes in it. I fit my material model to an ogden (n=3) strain-energy function using uniaxial tensile test from the lab. I am trying to recreate the stress/strain curve from the lab w/ two holes drilled through the middle of the tube. The tension of the tube is what is recorded in the lab tensile test. I am unable to see increased compliance of the tube no matter what I do.

I am having trouble figuring out which Output variables will give me the analogue of tension recorded in the lab. I originally thought that summing up the reaction forces (RF) on a nodal slice through the tube would give me the tension recorded in the lab. I see that the Reaction Forces are only calculated on the nodes where the displacement/load B.C. are applied. This makes it so there is no difference in reaction forces whether or not holes are present in the tube.

I also tried summing the Nodal Forces and Total Force (TF) over a nodal slice to give me the tension, and that didn't work either.

If anyone has an idea of a way to get my stress/strain curve of the tube w/ holes patterned into it to match the lab tensile test that would be great. I thought the easiest way to do it would be to get the tension in the tube (in abaqus) and divide by the engineering area to give the nominal stress of the tube.

Thanks A Lot,
David

Jorgen
2008-08-03, 20:12
What I would do is very similar to what you described: I would sum the reaction forces (RF) at one end of the tube (where you have BC or loads specified). You can sum and extract the total RF from within Abaqus/CAE. Then if you divide the total RF with the initial area you will get the engineering stress. I use this approach quite often and it works just fine.

- Jorgen

goldm066
2008-08-03, 20:25
Jorgen,


Thanks a lot for the swift response!

Yes, I do exactly as you described, although I am using CAE, so I field output the nodal reaction forces from the end face of the tube, and use a MATLAB script to sum them up over each time interval. The problem is that I get the same engineering stress, e.g. the same total reaction force on the end nodes regardless of whether or not there are holes in the tube.

In order to get the total strain of the tube, to match the tensile test, I am using the z-coordinate of the displaced face (displacing only z-direction), and calculate the change in length, and strains accordingly.

What I find very bothersome also, is that I can apply the same 1 million Pa load to the end face on the tube with and without holes, and I get nearly the exact same strains, no where near the lab results which reduce the nominal stresses by about 40% at 0.6 strain.


Thanks Again,
David

Jorgen
2008-08-03, 20:36
That sounds strange. Perhaps you can create a test case in which the holes are "really big". That way you should surely get a different response. If not, you will know that there is something else that is wrong (perhaps with the data extraction approach).

- Jorgen

goldm066
2008-08-03, 20:54
Jorgen,


I have tried exactly that, well similarly I tried patterning 6 holes instead of 2 and get no change in reaction force. This problem has been driving me crazy basically all summer as my job is to pattern holes with an optimal size and shape, and since I cannot match up a basic tensile test, no designing is getting done.


One of my professors suggested I use a fully linear elastic material model and try something similar, and if that works correctly, I would know it was the material model. Using that model, I tried using two holes, and was able to get an increase in compliance. Unfortunately I added 4 more holes to the fully linear model to see if it would further increase in compliance as it should, and it actually matched up exactly with the two holed fully linear model.

I am very new to FEM, and thought it was something wrong with the analysis; especially when I expected the reaction forces to be present in any cross-sectional slice, and they are only present on the B.C. nodes.

I can attach my input or CAE files once I get back to the lab.


Thanks A Lot for the Replies,
David

Jorgen
2008-08-03, 21:02
Still sounds strange. You should get the same reaction force at any cross-section. The reason I proposed the end sections with the BC nodes is that it is easier to extract the forces there. It is not necessary to use Matlab (although Matlab is really cool), you can actually use CAE to sum up the forces (which probably is faster).

If I were to guess, and this is just a guess, I suspect that you are doing something wrong when you extract the total reaction force...

- Jorgen

goldm066
2008-08-03, 21:10
Jorgen,

If you could please tell me the basics of how to have CAE sum up the reaction forces for me that would be excellent, as I do not know how to do this, and have tried looking online before.


You agree that there *should* be nodal reaction forces in all nodes of the tube? This must be the case in order for it to be in static equilibrium (right?), which makes since, just like the tension is constant everywhere in a piece of pulled string.

Today, I tried outputting the data from some middle nodes, just by selecting them with a box arbitrarily. All of the reaction forces were 0!



Thanks Again,
David

Jorgen
2008-08-03, 21:18
Well, there will be internal nodal forces during the simulation, but you cannot extract those very easily and they are not the same as the reaction forces. In finite element terms, "reaction forces" are only present at nodes at which there are a displacement BC applied. That is why you got zeros for internal nodes.

In short, in order to use CAE to get the total RF you need to first save the RF as a history output variable when running the simuation. Then you can plot the RF history variable within CAE. Simply select all RF terms for the element set that you are interested in and click "add", and you will get the total RF data.

- Jorgen

goldm066
2008-08-03, 21:32
Jorgen,


Very useful, I had been using field output, from the visualization module, then writing an X-Y report to a file, filtering the data in excel, then running it through my M-File, I will compare the two results.

So if the History output of the reaction forces you described matches my MATLAB output (I still get no increased compliance), I should assume then that there is something wrong with the software probably?

Thanks A Ton,
David

goldm066
2008-08-03, 23:07
Jorgen,

Unfortunately the history output-->sum command produced identical results as my MATLAB script.

Regardless of how many holes I cut in the tube summing the reaction forces on the end face and dividing by original area (engineering stress) returns my initial stress/strain curve used to define the material.

Any other advice you may have will be most useful. Is it possible something is goofy w/ our ABAQUS?


-David Goldman