Viscoelastic material model-am i taking the correct approach

[FYPNoddy]

Hi all,

I have been trying to gain an understanding of representing a viscoelastic material in abaqus. To do this i have been trying to represent polypropylene using data obtained from creep curves found in literature (attached). To now what i have been doing is the following:

In material manager: selecting Elastic-inputting Instantaneous Youngs Modulus, E, Poissons, v, (E=1.8GPa, v=0.4) and setting the moduli time scale to instantaneous. And Viscoelastic-Domain:Time, Time=Creep Test data and inputting Shear Test Data

Shear Test Data obtained as follows:

1. From the instantaneous youngs modulus, E and poisson's ratio, v, obtaining the instantenous shear modulus, G0=E/(2(1+v))

2. Picking a certain stress, and probing values of strain against time (from attached creep curve):
3. Calculating:
Creep modulus, E(t)=stress/strain(t)
Shear Modulus, G(t)=E(t)/(2(1+v))
Shear Compliance, Js(t)=1/G(t)
to finally obtain "Normalized Shear Compliance", js(t)=G0.Js(t) as requested by Abaqus. I guessed the "Long-term normalized shear compliance or modulus" based on the calculated data.

To test this inputted material property, I perform creep tests on a 1m x 0.5m model and visualise the results. With these results i try to recreate the creep curves.

The only curve that ever correlates is the curve i took values from. I was hoping for some guidance on the matter if anyone could provide it. Maybe i have been approaching the problem all wrong?

Thanks and Regards,
FYPNoddy

[Jorgen]

It seems like you are doing it right. Note, however, that it more difficult for Abaqus to handle creep data than stress relaxation data. It is possible that your experimental data is not covering a wide enough range of creep time to accurately determine the viscoelastic functions. It is also possible that a linear viscoelastic material model cannot fit your experimental data.

-Jorgen

[FYPNoddy]

Thanks Jorgen,

Yes i understand that Abaqus uses the creep data and converts it to relaxation data, through convolution integrals i think.
When you say the data is not covering a wide enough range to accurately determine the viscoelastic functions, could you explain this in a little more detail?
You are right that this linear viscoelastic material model does not fit the experimental data, not perfectly, but it does give a okay approximation for me. And I believe the only way to accurately model this non-linear viscoelastic material is with the use of a UMAT subroutine. Am i correct in sayin this? If so, could this be done with the use of these creep curves or would i need more data?

Thanks again for your help,
FYPNoddy

[artemioalanis]

Hi
I just stated working on a creep test and i found your post to be very usefull and i would like to thank you for that. But, there's allways a but in life
I can get my simulation to run. i get a ERROR 144
DO you know what it could be these are the data i extracted form your post
Stress Strain Time E(t) Creep G(t)=E(t)/(2(1+v)) Js(t)=1/G(t) js(t)=G0.Js(t)
(N/mm2) s
1 0
2000000 0.1 1000 20000000 7092198.582 0.000000141 0.000065
2000000 0.15 10000 13333333.33 4728132.388 2.115E-07 0.0000975
2000000 0.2 100000 10000000 3546099.291 0.000000282 0.00013
2000000 0.3 1000000 6666666.667 2364066.194 0.000000423 0.000195
2000000 0.5 10000000 4000000 1418439.716 0.000000705 0.000325

Could you help me get my simulation up and running

[FYPNoddy]

Hi, artemioalanis

First of all i want to say that the data in the .zip file is taken from Plastics Engineering book by R.J. Crawford, and i did a creep test on polypropylene specimen since that post and have found that the curves in the book are quite unrepresentative of my results. Also, polypropylene is a non-linear viscoelastic material (as are most polymers), and modelling viscoelasticity in Abaqus in the way outlined above only encompasses a linear viscoelastic response.

That said, looking at your results, it looks like your stress, strain, creep modulus, shear modulus and shear compliance are all good, but the normalized shear compliance values don't look so good, and these are the important ones as they are the ones that will eventually go into Abaqus..note that shear compliance must be multiplied by the instantaneous shear modulus, calculated taking the instantaneous value for Young's Modulus.

Attached is a link to the excel file I was using when i was getting this model up and running, hope it works:

http://dl.dropbox.com/u/3740620/Polypropylene1.xls

It found it interesting to see that the creep modulus can be represented as a function of time by a power law expression as shown.
Anyways, i think you might have been getting a 144 error due to your normalized shear compliance values, as i think those errors come about due to the program attempting to divide by 0, or values close to 0.

Hope this helps,
Regards,
FYPNoddy

[artemioalanis]

Thanks a lot for your support.

But abaqus wont convert the data and it says that value are to high and im with in range?
Did you have the same problem?

[FYPNoddy]

I had a few problems, but none like the ones you describe above. What are you running in your step module? It should be a visco step, and maybe run it for 10,000 seconds or so to get a good visual response.

Regards,
FYPNoddy