no-tension material in ansys

[SKJoe]

Hi

I would like to model soil continuum by plane42 in ansys, material model drucker-prager. but after initial solution I
realized that there are some tension stresses...Is there anyone who can help ?

I found this paragraph in book "The finite element methods Vol. 2" by Zienkiewicz O.C., Taylor R.L. (cited):

"it suffices to carry out the
analysis elastically and wherever tensile stresses develop to reduce these to zero. The
initial stress (modified Newton-Raphson) process here is natural and indeed was developed
in this context[102]. The steps of calculation are obvious but it is important to
remember that the principal tensile stresses have to be eliminated." - but how to eliminate this principal tensile stresses in ?

ps : I hope it is clear enough and sorry for my english

 

[EdR]

I am not familiar with Ansys and it's capabilities but typically the tensile stresses are eliminated by using a "tension cutoff" option of the material model...This forces tensile stresses to be reset to zero and redistributed throughout the model......

Ed.R.

 

[torpen]

Hello,

Why do you want suppress the initial stresses?

Material and/or boundary conditions can explain the origin of the initial stresses.

Firstable, you could test your model with a linear elastic material.

Then if the initial stresses are still present check out your boundary conditions.
You begin by deleting all these boundary conditions except those necessary to suppress rigid body modes.
And you add the other ones one by one to find the "guilty" one.
Or you guess the responsible boundary condition.

You can check the tolerance value of the residue of the iterative process (numerical aspect).


Regards,

Torpen

 

[GBor]

Not sure about the specific coefficients for Drucker-Prager, but if you can use a Von Mises Curve or similiar material model, set the stress strain curve so that when strain is positive (tension), the curve flattens (zero modulus, so no load carrying capability). If the strain is negative(compression), give it an appropriate modulus to carry load. stress = modulus * strain...if modulus is zero, so is stress.

 

[EdR]

GBor:

You have to check how the specific code (and material model) handles a zero modulus....In most codes a zero modulus will give a singular stiffness matrix and the solution will blow up...Depending on the model, using even a modulus close to zero will also cause problems.....

Ed.R.

 

[torpen]

Hello,

The problem could be due to the friction angle parameter of the drucker prager model.
If this parameter is greater than a critical value the criterion accepts tensile stresses.

Regards,

Torpen.