Learning Forum

[Giorgos Troulliotis]

Dear all,

I am working on cardiac simulations and I have implemented a subroutine which couples a custom 0D CV model to an FEM model of the heart. The segregated coupling is centered around minimising a residual function. 

I would like for the implicit solution to move to the next timestep once the coupling residuals have reached a certain threshold in addition to the standard displacement and energy residuals converging.

Currently, I am forcing convergence after a fixed amount of iterations. This produces reasonable convergence, however it's quite inefficient and the degree of convergence varies during the cardiac cycle. 

Is there any way to create a custom convergence criterion, or control solver convergence outside of the keyword *CONTROL_IMPLICIT_SOLUTION?

Many Thanks     

[Reno Genest]

Hello George,

If I understand correctly, currently, you are using the default implicit convergence criteria in LS-DYNA and force convergence after a fixed amount of iterations correct? Then, you check your residual function and it is low enough after the fixed amount of iteration so that you can say convergence was reached correct? But, you would like to implement a custom convergence criteria based on your residual functions? Do I understand correctly?

 

Reno.

[Giorgos Troulliotis]

Hi Reno,

That is correct I have found that 10-15 iterations works for most timesteps in the cycle however, some timesteps require less and some more. I would like to automate the process for efficient solution and control over the coupling convergence.

Best Regards,

George

   

[Reno Genest]

Hello George,

Ok, let me check with the developer and get back to you.

 

Reno.

[Giorgos Troulliotis]

Hi Reno,

Have you had heard back form the developers regarding this query?

Best Regards,

George

[Reno Genest]

Hello George,

I haven't heard yet. I just followed up with him. I'll let you know when I hear from him.

 

Reno.