nonlinear optmization and identification hyperelastic parameters

**sanchez1312** · 2012-06-05, 13:18

I am doing nonlinear fitting of mooney-rivlin by help of Matlab using lsqcurvefit, or lsqnonlin,but i can fit only separeted curve not 3 curves from tensile test,pure shear and compression
i have read that if i can describe multi-axial loading i need more than tensile test to fit,but i dont know how can i do fit or optimization of parameters simultanously to get optimized parameters which will be suitable for mult axial loading

Can anybody help me?

thank you

**IceBreaker** · 2012-06-05, 16:21

HEEDS MDO, MCalibration, may be even the super cost-effective (academic version, at least) Hyperfit should be able to do what you are trying to accomplish using your own code - fast and easy.

**Jorgen** · 2012-06-05, 19:03

What FE program are you using? The Mooney-Rivlin model can be calibrated by, for example, both Abaqus and ANSYS.

Also, what material are you interested in? I rarely use or recommend the Mooney-Rivlin (MR) model. There are other material models that are equally accurate, easier to calibrate, and more robust.

I might be a bit biased on this, but the MCalibration application is absolutely fabulous for material model calibrations. If you haven't you should check it out.

-Jorgen

CF · 2012-06-11, 10:49

Presumably, you have set up an objective function for each individual test. You need to set up one objective function that contains data from all three test modes. For example, say you have a tensile, compression, and pure shear test data with 100 points in each set. After you run the simulation/analysis for each test mode, you calculate your objective function. The first 100 points in your function will be the difference between your tensile model and tensile experimental results; the second 100 points in your function will be the difference between your compression model and compression experimental results; and the last 100 points in your function will be the difference between your pure shear model and pure shear experimental results

Fobj = [ (Tensile_Model_Results - Tensile_Exp_Results) (Compression_Model_Results - Compression_Exp_Results) (Pure_Shear_Model_Results - Pure_Shear_Exp_Results) ]

Fobj is an array with 300 points and this is returned to the lsqnonlin function. In your case, you may have some weighting in your objective function. That depends on your needs.

Hope that helps.

Cormac