I wrote a damage model in UMAT subroutine based on "Probabilistic damage model for the dynamic fragmentation in brittle materials",Hild,Advances in applied mechanics,pp.1-72,2010;Multiscale model,2002,Hild.The procedure I followed in UMAT subroutine is as follows:
1.I calculate the principal stresses.
2.Randomly generate a failure stress according to equation(84) in 2010 paper.
3.For each principal stress I calculate the modified defect growth density given in equation(83) in 2010 paper.
4. Corresponding to each principal stress a damage value is found according to the conditions given in equation(81) in 2010 paper(NOTE:Anisotropic damage model is used in three orthogonal directions D1,D2,D3 )
5.I update the damage tensor/compliance tensor using (33)(JPEG file is attached)as provided in multi-scale model paper(2002).
6.I update the stress and repeat the steps:1-6.
NOTE:1.The material is isotropic elastic needs only E,poisson's ratio,material parameters.The other material parameters like weibull parameters are also given. Refer the papers provided for clarification.
2.I am dealing with Impact problems so I need VUMAT
I have a doubt in VUMAT if I calculate compliance tensor(as show in attached file) at each step and inverse it to obtain the constitutive tensor and then multiply with strainInc to obtain stress increment.I will not update Jacobian.Can I do this in VUMAT to calculate stressNew?.
The DOUBT I have in VUMAT subroutine coding is how to update the stress according to the constitutive law as Jacobian matrix is not updated in VUMAT to calculate stress,How to update stress.Please detail me the procedure in VUMAT.Do I need Rotation matrix to rotate the stressold, strainInc array in VUMAT to calculate updated stress?.Do I need stress rate like Green-Naghdi stress rate in VUMAT?.Please help me by answering the question.Thank you
I attach file of the subroutine in UMAT
Hi Bhanu,
In a VUMAT, Abaqus provides you with a 'straininc' (incremental strain) and expects from you a 'stressinc' (stress increment). Before entering the subroutine, Abaqus already does take account of the rigid rotation for you and after you update your stress, it rotates it back to global frame to calculate nodal force. You do not have to provide any rotation.
Secondly, a VUMAT is used in explicit dynamic analysis. So, you don't need to use a jacobian unlike implicit . In explicit, the time increments are very small, hence a jacobian is not needed. For your anisotropic damage, please formulate different damage variables corresponding to different material directions (based o your physical principle) and degrade the effective stresses using those damage variables. This will induce anisotropy in damage.
-Regards,
Supratik
Hi Bhanu,
In a VUMAT, Abaqus provides you with a 'straininc' (incremental strain) and expects from you a 'stressinc' (stress increment). Before entering the subroutine, Abaqus already does take account of the rigid rotation for you and after you update your stress, it rotates it back to global frame to calculate nodal force. You do not have to provide any rotation.
Secondly, a VUMAT is used in explicit dynamic analysis. So, you don't need to use a jacobian unlike implicit . In explicit, the time increments are very small, hence a jacobian is not needed. For your anisotropic damage, please formulate different damage variables corresponding to different material directions (based o your physical principle) and degrade the effective stresses using those damage variables. This will induce anisotropy in damage.
-Regards,
Supratik
Hi Bhanu,
In VUMAT you dont need rotation matrix.
Or as per the paper sylvain shared, you can use the storage scheme of VUMAT and UMAT to convert an UMAT program into VUMAT.
you can find the interface between umat (Abaqus/Standard) and vumat (Abaqus/Explicit). in the following link:
http://www.soilmodels.info/
Please check the following link for wrting VUMAT :
http://imechanica.org/files/Writing%20User%20Subroutines%20with%20ABAQUS.pdf
Best regards,
Sairam
- Badges
- Science topic