About the issue of solver-artificial-viscosity

Just Starting
1

Hello everyone,

I would like to ask about the issue of solver-artificial-viscosity,

solver-artificial-viscosity
max-artvisc = 0.01
s0 = 0.01
kappa = 5.0 ; 10.0

For example, I saw a brief description of the specific meanings of these three parameters in the help file, but I don’t quite understand their physical meanings. For example, what determines the size of max artvisual and how does s0 determine the truncation value?

Will the value of max artvisual be too large or too small, resulting in distorted results? I am very confused about these.

Perhaps this question is a bit foolish, but I hope to get everyone’s answer.

Best regards.

2

This artificial viscosity method is based on that of Per-Olof Persson. The paper is the best source of information: http://persson.berkeley.edu/pub/persson13transient_shocks.pdf

3

It’s very useful, thank you for your help.

I have another question to consult with you. In this article, it is mentioned that ‘k are chosen empirically sufficiently large so as to obtain a sharp but smooth shock profile’.

So if I specify a large kappa in ini file, such as 50, then the effects of s0 and max artvisual will be minimal. I understand that in this case, capturing shock waves has strong robustness, but the computational cost will be greatly increased, right?

Best regards.

4

So far as computational cost is concerned it partially depends on the amount of overall viscosity added. More viscosity results in a lower CFL limit for the element that the viscosity is being added to. Now, if that element is large this will not impact the overall time step (which is set by the lowest CFL in the entire domain). However, if the element which the viscosity is being added to is already quite small then this may cause the time step to drop.

Regards, Freddie.

5

Thank you for your patient explanation.

Good wishes.