preCICE adapter for MBDyn solver for beam elements

[Claudio]

Hi @Makis ,
of course I will. I have just concluded some simulations and I’d like to share some of the results. As I first wanted to validate the adapter, I focused on the Turek-Hron benchmark which is one of the tutorials (this is named FSI3 in the paper). At the moment I am still not able to overcome instability due to mass ratio (solid/fluid) close to 1.
So I decided to consider the other benchmark (named FSI2) which has a mass ratio of 10. My results, considering the tip movement, are the following:

The y displacement has a mean value of 3.35mm with an amplitude 83.9mm and a period of 586ms. The reference has a mean of 1.23mm with an amplitude of 80.6mm.
The x displacement has a mean value of -14.54mm and an amplitude of 11.65mm and a period of 293ms. The reference has a mean -14.56mm and an amplitude of 12.44mm.
I’ve got also other data but I’d like to keep the focus of the post on the configuration aspects and open points:

• I need to use a progression of forces: the MBDyn mesh is the same as the PreCICE Solid mesh. The forces on the PreCICE nodes coming form the coupling are passed to MBDyn multiplied by a coefficient:

the first 500ms are used to let the fluid settle, and then for the next 500ms the forces are increased. The slope of the coefficient is really critical for convergence. A steeper slope makes the computation unstable. But this way the simulation requires more time.

• I have been using the following coupling parameters:

    <coupling-scheme:serial-implicit>
    	<time-window-size value="1.e-3" />
    	<max-time value="20."/>
    	<participants first="Solid" second="Fluid"/>
    	<exchange data="Forces0" mesh="Solid-Mesh" from="Fluid" to="Solid" initialize="0"/>
    	<exchange data="Displacements0" mesh="Solid-Mesh" from="Solid" to="Fluid" initialize="0"/>
    	<max-iterations value="50"/>
    	<relative-convergence-measure limit="1e-4" data="Displacements0" mesh="Solid-Mesh"/>
    	<relative-convergence-measure limit="1e-3" data="Forces0" mesh="Solid-Mesh"/>
    	<acceleration:IQN-ILS>
    		<data name="Forces0" mesh="Solid-Mesh"/>
    		<preconditioner type="residual-sum"/>
    		<filter type="QR1" limit="1e-2"/>
    		<initial-relaxation value="0.1"/> 
    		<max-used-iterations value="100"/>
    		<time-windows-reused value="10"/> 
    	</acceleration:IQN-ILS>
    </coupling-scheme:serial-implicit>
  

1. The convergence measures require quite a lot of iterations to converge:
   

   

2. I can only use Fluid as second participant beacuse otherwise I have issues. But I would be interested in trying other configurations.

I would be interested in discussing some ways to speed-up the convergence and finding a way to overcome the issues regarding the mass ratio.
I am going to update this post when I have new findings
best regards
Claudio