【讨论】请教个问题 , 欢迎cafe0519进来. About UMAT

meyh

真诚的向大虾请教.
  
When writing a UMAT, if you lwant to input a material parameters, which   is not a constant, instead it is a fuction of stress and strain, How can we input it into a UMAT?
  
Bow!

wjytj

define them as SDVs (solution-dependent state variables) in UMAT, not as constants given in the inp file.

meyh

Thanks a lot. It's a good suggestion.I 'll try this and let you know the results.
  
You guy seems to be an expert on ABAQUS.

cafe0519

meyh wrote:
You guy seems to be an expert on ABAQUS.

  
maybe, you havenot known dear bro. wjytj is an expert of abaqus.
He finished his doctor degree by successfully using UMAT.
  
btw, I am not familiar with UMAT.

meyh

You are modest. I read your post for others.Anyway thanks for providing this information.So I should discuss more with wjytj.

meyh

wjytj, I set them as SDV but it seems that there is still some problem with it. Because the UMAT can't be run and there is no error message either.
  
I know,  and cafe0519 also said, you are an expert on UMAT, could you give me more hint about it? I derive my own constitutive equations and trying to put them in UMAT. As I said before in the constitutive equation there are some material parameters which are fuctions of stress and strain.And these parameters are used to calculate the stress and strain. This make me a little bit confused. Which I should caculate first, the paprmeters or the stress?
  
I am really appreciated you patient and you reply. If you are interested in this you can email me at
meyh2002@yahoo.com.
  
Thanks a lot.

wjytj

I also think that Bro. cafe is so modest
  
In UMAT, you will update the stress, the Jacobi matrix and the SDVs, and  
don't forget that the Newton-Raphson numerical algorithm is used to solve  
the nonlinear equations, so the order of which is calculate first is really not
important as you imagine, since at last all the updated values have to be
converged.
  
In spite of the above fact, I suggest that you could update SDVs first  
since they are functions of strains, then update stress and DDSDDE  
using the calculated SDVs, at last update the SDVs again using the
calculated stress.
  
Of course, it is your own duty to ensure the correctness of your constitutive  
relations and the corresponding numerical algorithm.

meyh

Your explanation is very reasonable and clear. The problem is the parameter is also a function of stress. There should be an initial vale for stress so that the parameter can be calculated.In my model the initial strain can be regarded as zero, can I also regard the stress to be zero?
  
BY the way, if I regarded the parameters as constants I could get reaonable outputs. Based on this I think the constitutive equation and  the corresponding numerical algorithm are correct. I think the problem is in my UMAT.
  
I will try the UMAT again.
  
THanks a lot !