How to recreate NUL...
 
Notifications
Clear all

How to recreate NULL material model with dynamic viscosity term in LS-Dyna using a UMAT?

6 Posts
2 Users
0 Likes
684 Views
Posts: 3
Topic starter
(@kennedy)
Barista
Joined: 4 years ago
I am trying to duplicate the LS-Dyna NULL material model with the dynamic viscosity term using a UMAT.
As per the LS-Dyna manual, when using the NULL material model with dynamic viscosity, a deviatoric stress term is added using the following:
S_ij = 2*mu*e'_ij
where S_ij is the stress tensor, mu is the dynamic viscosity, and e'ij is the deviatoric strain rate.
I have implemented the UMAT as follows:
davg=(eps(1)+eps(2)+eps(3))/3.
savg=(sig(1)+sig(2)+sig(3))/3.
sig(1) = sig(1) - savg + 2.*mu*(eps(1)-davg)/(dt1)
sig(2) = sig(2) - savg + 2.*mu*(eps(2)-davg)/(dt1)
sig(3) = sig(3) - savg + 2.*mu*(eps(3)-davg)/(dt1)
sig(4) = sig(4) + 2.*mu*eps(4)/(dt1)
sig(5) = sig(5) + 2.*mu*eps(5)/(dt1)
sig(6) = sig(6) + 2.*mu*eps(6)/(dt1)
When I compare the results against the built in NULL material model, the results match fine when the dynamic viscosity is 0, but I can't get the results to match at all when I assign a value to the dynamic viscosity.
Does anyone know where I am going wrong?
5 Replies




Negative Volume
Posts: 639
Admin
(@negativevolume)
CEO
Joined: 5 years ago

Hi @kennedy,

I apologize for missing this topic. Seems like an interesting venture to undertake. I'd be interested if you figured out what the issue was. 

Reply
4 Replies
(@kennedy)
Joined: 4 years ago

Barista
Posts: 3

I still haven't solved this issue. I've narrowed it down a bit in that it looks like the strain rates that I am calculating (using eps/dt1) do not match the strain rates reported from LS-Dyna when using the built in Null material model.

Reply
Negative Volume
Admin
(@negativevolume)
Joined: 5 years ago

CEO
Posts: 639

@kennedy

Posted by: @kennedy

the results match fine when the dynamic viscosity is 0, but I can't get the results to match at all when I assign a value to the dynamic viscosity.

What are you trying to match to when you assign a dynamic viscosity? I would not think it would match the built-in *Mat_null material because that was the point of re-creating the null material but with dynamic viscosity. Maybe I am just not sure what your goal of this is. 

Reply
(@kennedy)
Joined: 4 years ago

Barista
Posts: 3

@negativevolume, My goal is actually to modify this null material model eventually, but to start I wanted to make sure that I could replicate the built in one first. The built in null model has a dynamic viscosity term already. So the answers should be very similar if not exactly the same. They are good when you ignore the dynamic viscosity term, but this just means that the hydrostatic portion is good, but the deviatoric portion is not right. The deviatoric portion is controlled by the value of dynamic viscosity (which I make the same for both) and by the strain rate (which is what the difference is from my investigation). I just can't figure out how I should be calculating the strain rate. The LS-Dyna theory shows it as simply the strain increment divided by the timestep, which is what I am doing. Perhaps it is actually calculated in a different manner?

Reply
Negative Volume
Admin
(@negativevolume)
Joined: 5 years ago

CEO
Posts: 639

@kennedy Okay, I understand now. This is probably obvious, but what type of element are you using to test the materials? I'm pretty sure shell and beam elements are not affected using *Mat_null (except for contacts) so you should be using solid elements. 

Ah, also make sure you know how *Mat_null is calculating strain (true vs engineering). I know LS-Dyna outputs strain as True Strain. This might be where you are getting your difference between the material models. 

https://www.dynasupport.com/howtos/general/strain-rate

Reply







Share: