[Joachim]

Hey everyone!

I am using the Jameson dissipation terms in my solver, and I was wondering what to do at the boundaries. I read in some papers that the dissipation was sometime set to zero at the wall. Why is that so? why shouldn't I compute the value of the dissipation term at the wall assuming that u = 0 (viscous)?

I guess that at the inlet/outlet, I can simply compute the dissipation terms from the boundary values.

Any suggestions??

Thanks!

Joachim

[FMDenaro]

Originally Posted by Joachim

Hey everyone!

I am using the Jameson dissipation terms in my solver, and I was wondering what to do at the boundaries. I read in some papers that the dissipation was sometime set to zero at the wall. Why is that so? why shouldn't I compute the value of the dissipation term at the wall assuming that u = 0 (viscous)?

I guess that at the inlet/outlet, I can simply compute the dissipation terms from the boundary values.

Any suggestions??

Thanks!

Joachim

artificial dissipation is just a trick to stabilize the numerical solution across zones where strong gradients are present and can generate oscillations, so it should be added only locally

[Joachim]

Thanks for your answer!

However, I am using a central scheme for my convective terms (+RK order 4), so that I need fourth order artificial viscosity to remove any odd/even decoupling.

[FMDenaro]

Originally Posted by Joachim

Thanks for your answer!

However, I am using a central scheme for my convective terms (+RK order 4), so that I need fourth order artificial viscosity to remove any odd/even decoupling.

but artificial dissipation is modulated by the local gradient intensity... do you have strong gradients as BC.s?

[Joachim]

well, I am computing the flow over a laminar flat plate right now, so I have gradients at the wall...

[FMDenaro]

Originally Posted by Joachim

well, I am computing the flow over a laminar flat plate right now, so I have gradients at the wall...

This is not clear ... when you solve laminar flow, there is no unresolvable gradient in such case, therefore the artificial dissipation does not make sense...

[Joachim]

I have the values at time step n, and I explicitly compute the fluxes to get the state vector at time step n+1. However, I don't know if I should include the artificial viscosity at the wall or not...