Modeling a thermomechanical bending process 1

[Oueg18]

Hi,
I'm writing on this forum because I'm a bit at a lost with an analysis conducted with ANSYS workbench.

I've got creep response data and springback data: I used a test bench consisting of a constant weight pulling laterally a polypropylene flat bar (somewhat in the manner of a cantilever beam but pulled honrizontally)
SEE PICTURE in attachment Schematic and creep response equation.pdf
For the creep test: The heat gun warms the flat bar while the force is pulling on it: the magnitude of the angular deformation is measured as a function of time.
For the springback test: The heat from the heat gun and the force are both removed and the springback is measured as a function of time.
I've developped a creep equation for the creep test that also figures in the ATTACHMENT. Schematic and creep response equation.pdf
With matlab, I've fitted the equation and found the right parameters to fit my data.

I am now trying to model this experiment in ANSYS workbench. I've read a bunch on the subject but I'm having a hard time finding where to start and what should be included in my analysis.
From what I understand, I need to make a static structural analysis with the right temperature profile. That I can do. But so far, my analysis is carried out only in the elastic domain and this won't work since a permanent deformation remains in the flat bar, after the springback.

I know I must include PLASTICITY and/or VISCOELASTICITY in my material properties (like MISO and prony). But workbench asks for stress VS strain data and shear modulus VS time data. All I have is series of data of angular deformation (in rad) as a function of time (seconds).

Does the analysis must include both MISO (multilinear isotropic hardening) in the PLASTICITY TAB and viscoelastic shearing in the VISCOELASTICITY TAB? Is viscoelastic shearing sufficient alone? Because if both are needed I could go to my university and make some tensile test to obtain Stress VS strain data.


How can I use my results in Workbench? Can I include my creep response equation in Workbench?

How should I make an FEA analysis of this experiment using Ansys Workbench? I'm a bit familiar with command snippets in mechanical but I know next to nothing of mechanical APDL (Ansys classic)

Please help me, my director knows nothing in ANSYS and nobody at my university either

 

[Oueg18]

The attachment picture[URL unfurl="true"]https://res.cloudinary.com/engineering-com/image/upload/v1481140477/tips/Schematic_and_creep_response_equation_jvobkd.pdf[/url]

 

[Oueg18]

Anyone can help me please?

 

[rickfischer51]

First, its not clear to me why you want to include viscoelasticity. If you ran the analysis at the same strain rate as your test, you would see no difference due to visco effects. Of course you would have to run it as a transient dynamic rather than a static analysis. I would run this first without the visco. You can add that later.

As far as I know, if you want to use your own creep law you need to incorporate a user material, which means recompiling and relinking the program. This is not trivial, and you would want this version installed on a computer that others cant access, because it will mess up their results. Try and use one of the standard creep models. The documentation describes them as used for primary or secondary creep, but ignore that. Creep models are simply a curve fitting exercise, and the time constant is a function of your test data. IF the equation fits your data, it doesn't matter what other people use it for.

I do not believe you can use you data to characterize you material. You have load and rotation on the tip of a cantilever beam, and stress and strain with vary through the length and thickness. The best you could do is try to integrate your results over the beam and get a representative average value, but you will be combining tensile and compressive effects. Creep will be different in tension and compression because free volume and hence chain mobility will be larger in a tensile field than a compressive field.

In general, it is good to start slow and add complexity only when you are confident in what you have done up to that point. You seem to want to jump into the deep end, and it doesn't sound like you have much FEA experience. You do not want to be messing around with a user material law, multiple material nonlinearities, etc, until you have mastered the basics.

Rick Fischer
Principal Engineer
Argonne National Laboratory

 

[Oueg18]

I know I still got lot to learn about ANSYS.

Would you be so kind as to send me sequential steps to follow? like 1. ... 2. ... 3. ... and so on

Maybe I should have said that the flat bar, once the force is removed, springs back towards it's original position and this, during 5 to 6 minutes.

Here is an set of data (Those are degrees): Time is in minutes when the force is on and time is in seconds when the force is removed.

Please can you indicate what should be added to the material in WORKBENCH by referring to the related tabs in "Material data" (Such as the "Plasticity" Tab, the "Creep tab" and the "Viscoelasticity" tab)

 

[rickfischer51]

No, I am not going to do your homework for you I had to do my own, and so should you.

What you are calling spring-back is not spring-back. It is viscous recovery. Spring-back is instantaneous, and the result of yielded and un-yielded material fighting to establish static equilibrium after the load is released. Also, if you are looking to model this viscous recovery phase, then you do need a visco material model. Also also, plastic flow in a polymeric glass is sometimes a function of the hydrostatic stress component, so a yield model like extended Drucker-Prager may be more appropriate than MISO, although I think this evvect is not as pronounced in semicrystaline materials as it is in amorphous grades.

Rick Fischer
Principal Engineer
Argonne National Laboratory

 

[Oueg18]

Thanks for your answers, it is much appreciated.

My Masters director came to see me today and was advising me to post these pictures.

 

[Oueg18]

The angular deformation at

lim = 4.5 Degrees
t -> ∞

Question #1:

I guess I should start with a standard model before I even start thinking of using my own creep law.

I can't find Extended Drucker-Prager model in any of the tabs in the workbench material toolbox. From what I understand, I must use Extended Drucker-Prager material model via command snippets in mechanical. Considering my experimental results, which form of the EDP model would you advise to use, linear, power law or hyperbolic? And how would you go in coding the command to define the model?

MP,GXY or ALPD ?,MAT,C0?
MP,NUXY,MAT,C0?
TB,EDP,1
TBDATA,1,2.9,32,0 ???

I don't know how to compare my experimental curve with the model proposed by ANSYS once it has been defined. How can I see if the model fits my data?

Also, would that model be applicable to both parts of the experiment? (before AND after the removal of the force)

Question #2:
Do I have to include another model besides EDP? Such as a standard plasticity model in the "Plasticity Tab" from the workbench material toolbox? Or a standard viscoelastic model in the "Viscoelastic Tab" from the workbench material toolbox? Or maybe another model altogether via a command snippet? I'm just having a hard time figuring what type of non-linear material property has and doesn't have to be included in this experiment in order to get good numerical approximations.

Thanks a lot for everything, you're a great help