Search
Close this search box.
Notifications
Clear all

Hysteresis in ABAQUS(rubber indentation)

13 Posts
4 Users
0 Likes
1,636 Views
Posts: 3
(@gabriele77)
New Member
Joined: 14 years ago

Help! for the love of all that is good.

So after two weeks of struggling with this model, it took this long to figure out that hyperelastic materials are not valid for 3D elements in ABAQUS. I got some default experimental data for the rubber and I used it in this model. All lengths are in mm since abaqus defaults to .04 length when you input .045 for a radius. Therefore all material parameters and loads are in N/mm^2 or MPa. However, in spite of all this, I am still having convergence issues even with a load as small as 500 KPa. The input file is attached. Please, if you look at the input file and run it or anything like that, post something about it, anything. Otherwise, what would be the point to look at it and not posting a thing. Anything would help me right now because i am stumped. Im getting excessive distortion errors so the only thing I can think is that the load is too big which makes no sense or the mesh is not fine enough. Thanks.

Reply
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

Your simulation file crashes right away. I dont have the time to fix it, but I suspect that your definition of the contact or boundary conditions are not correct 🙁

-Jorgen

Reply
Posts: 3
(@gabriele77)
New Member
Joined: 14 years ago

Thanks

Jorgen,

Thanks a ton for some feedback. I used Tie contact/constraint with the four different parts in the model. Tie between the steel press block and alluminum sleeve, tie between the alluminum sleeve and the rubber bushing, and tie between the rubber bush and the steel shaft in the middle. I chose this because i usually find tie contact/constraint to be the most hassle free. As for the BCs, Im unsure as to what ones should be used for a 2D model since im used to 3D. As a result, I usually pick encastre or pinned. I can see where there would be trouble in choosing this bc since I am specifying a 0 DOF in the direction that is already defined to be zero. Maybe just U1 and U2 should be set to zero and thats all. Let me know what you think when you are free. Thanks!

Reply
Posts: 7
(@xthuy)
Active Member
Joined: 14 years ago

Dear Members,

Although ABAQUS has a very neat regression tool for the hyperelastic models, I havent found anything to do the regression for the S,m,C and A parameters for the Hysteresis sub-option.
I was wondering if anyone has any suggestions to do this fit using Creep Data (Tangent Modulus vs Time for different stress levels).

Thanks in advance,

Joaquin

Reply
Posts: 3998
(@jorgen)
Member
Joined: 5 years ago

Great question.

I have not announced this much yet, but I have developed a material model calibration software (called MCalibration) that can be used to calibrate the [FONT=Courier New]*Hysteresis[/FONT] model, and a number of other material models. If you are interested, send me a private message and I will give you a free 1-month trial license to this software.

I will create a nice announcement and demonstration of how this software can be used shortly...

-Jorgen

PS. Yes, the MCalibration application can use creep data to determine the S, m, C, and A parameters.

Reply
Posts: 7
(@xthuy)
Active Member
Joined: 14 years ago

Dr. Bergstrom,

I have a limited amount of test data from the manufacturer. Ive attached the Creep Modulus vs Time curves at different stress levels. I understand that these test are uniaxial - tension and that the stresses are nominal. From this curves, I can get the creep strain vs time at each stress.

I am wondering if this data is enough to determine the coefficients associated with Network B (Hysteresis Sub Option in Abaqus). In the Section 5.4 of your PhD Dissertation, a procedure for the determination of these coefficients is explained. It seems that at least two tests done at different strain rates and including loading and unloading are required.

My concern is if the data that I have available will be enough to capture the rate dependency of the model.

Just in case, I would be using Mooney-Rivlin as the hyperelastic model.

Thanks,

Joaquin Gutierrez

Reply
Page 2 / 2
Share: