Effect of interface type7/type2 on nodal time step

Gabriela G

Hello everyone

I am currently comparing different FE models and their effect on computational time. My models consist of simplified screws (rigid bodies)  inserted into a bony cylinder. I have two parameters, which I change in my models: E-modulus and interface type. Apart from this, the models are always identical. I have compared the following three models:

- type 2 interface, low E-modulus (Model 1)
- type 2 interface, high E-modulus (Model 2)
- type 7 interface, high E-modulus (Model 3)

The simulation time for model 2 is about seven times longer than for model 1 and three times longer than for model 3.
The nodal time step for the models with type 2 interfaces is always determined by the master node of the rigid body.
I run the models in RADIOSS.

Now, I have the following questions:

1. Why does an increase in E-modulus lead to such a high increase in simulation time (or decrease of the nodal time step) for the models with type 2 interface?

2. What is the difference in the calculation of the nodal time step for type 2 and type 7 interfaces?
And why is the nodal time step higher for the model with the type 2 interface when compared to the same model with a type 7 interface?

Thank you already in advance for your help.
Best regards,

Gabriela

Mathis

Hi Gabriela,

1. Why does an increase in E-modulus lead to such a high increase in simulation time (or decrease of the nodal time step) for the models with type 2 interface?

==> That's because time step (elemental or nodal) is directly linked to the value of the stiffness (stiffness which is initially driven by the young modulus).

You can find the formula of the time step in the Radioss user guide : https://2020.help.altair.com/2020/hwsolvers/rad/topics/solvers/rad/time_step_nodal_time_step_control_r.htm

You can see that the higher the stifness is, the lower your nodal time step will be.

____________________________________

2. What is the difference in the calculation of the nodal time step for type 2 and type 7 interfaces?
And why is the nodal time step higher for the model with the type 2 interface when compared to the same model with a type 7 interface?

==> Is the behavior the same between type2 & type7 model? The 2 interfaces don't have the same purpose.

 

Thanks & Regards,

Mathis.

Gabriela G

Mathis Loverini_20748 said:

Hi Gabriela,

1. Why does an increase in E-modulus lead to such a high increase in simulation time (or decrease of the nodal time step) for the models with type 2 interface?

==> That's because time step (elemental or nodal) is directly linked to the value of the stiffness (stiffness which is initially driven by the young modulus).

You can find the formula of the time step in the Radioss user guide : https://2020.help.altair.com/2020/hwsolvers/rad/topics/solvers/rad/time_step_nodal_time_step_control_r.htm

You can see that the higher the stifness is, the lower your nodal time step will be.

____________________________________

2. What is the difference in the calculation of the nodal time step for type 2 and type 7 interfaces?
And why is the nodal time step higher for the model with the type 2 interface when compared to the same model with a type 7 interface?

==> Is the behavior the same between type2 & type7 model? The 2 interfaces don't have the same purpose.

 

Thanks & Regards,

Mathis.

Dear Mathis

Thank you very much for your quick reply!

 

Ad 1:
This makes sense to me. However, I have also been running other simulations, where this explanation does not fit. 

- type 7 interface, low E modulus (model 4)
- type 7 interface, high E modulus (model 5)

With these simulations, model 5 is actually running faster than model 4. In model 4 the rigid body displacement is much higher than in model 5. From this observation I assume, that there is more deformation in the surrounding material in model 4. More deformation might lead to smaller elements, which would reduce the time step. 

Is it possible that, even though model 5 is stiffer and should have the smaller nodal time step, model 4 undergoes more deformation, leading to smaller elements, which outweights the effect of the E-modulus and thus, model 4 is running even slower?
What do you think about this?
Or is there another explanation for this behaviour?

 

Ad 2:

No, the two models behave differently. The type7 model shows much more movement when compared to the type2 model. I believe this makes sense, since type 2 is a tied interface while type 7 models fricition.
I just don't really understand why the choice of interface has such a big influence on the time step calculation.