[mb.pejvak]

Hi guys;
I applied a source term on the Non-zero pressure gradient flow over flat plate. the flow experiences natural transition to turbulence. The code worked for the flow without source term reasonably. By applying source term, the code diverges.
the source term is synthetic jet on flat plate and not function of independent variables, hence they don't change during iterations. it has strong component toward -Y. by elimination of the Y component, the code works well, but after applying Y component, it diverges.
I used slightly small under-relaxation numbers and different convetion discritization schemes (MUSCL, QUICK, TVD (mid-mode, superbee ,..)), but they did not make a big change (some of them postpone divergence).
I want to know whether or not there is method can be used to handle this kind of source term.

[FMDenaro]

Originally Posted by mb.pejvak

Hi guys;
I applied a source term on the Non-zero pressure gradient flow over flat plate. the flow experiences natural transition to turbulence. The code worked for the flow without source term reasonably. By applying source term, the code diverges.
the source term is synthetic jet on flat plate and not function of independent variables, hence they don't change during iterations. it has strong component toward -Y. by elimination of the Y component, the code works well, but after applying Y component, it diverges.
I used slightly small under-relaxation numbers and different convetion discritization schemes (MUSCL, QUICK, TVD (mid-mode, superbee ,..)), but they did not make a big change (some of them postpone divergence).
I want to know whether or not there is method can be used to handle this kind of source term.

The presence of a source term affects the stability constraint, have you checked for smaller parameters?
Are you using DNS/LES/URANS?
Furthermore, I suspect that it must be enforced in such a way to fulfill the continuity equation (divergence-free velocity)

[mb.pejvak]

Originally Posted by FMDenaro

The presence of a source term affects the stability constraint, have you checked for smaller parameters?
Are you using DNS/LES/URANS?
Furthermore, I suspect that it must be enforced in such a way to fulfill the continuity equation (divergence-free velocity)

Thank you Filippo for your quick reply. I also used smaller under-relaxation (about 0.01), but it did not make any change in convergence, it only slowed down divergence pace.
I used SST K-Omega (RANS) turbulence model. the developed code was evaluated for the case without source terms. I developed the code to simulate NZPG flow over a flat plate.
And about your third statements (fulfilling the continuity equation), I can not understand what exactly you mean and what I should do to reach it. could you please explain more.

[FMDenaro]

I mean, is your source term in the momentum equation divergence-free?
Otherwise it must be included as source in the pressure equation

[mb.pejvak]

Originally Posted by FMDenaro

I mean, is your source term in the momentum equation divergence-free?
Otherwise it must be included as source in the pressure equation

To illustrate, I used SIMPLER method, MIM in a collocated grid. The continuity equation by SIMPLER method is changed to the equation for pressure.
the source term (included in "b" in momentum equations) impose on the momentum equation and indirectly (in U_hat and V_hat) affects the pressure equation. But as far as I know there is not any term in pressure equation on which source terms can be applied. Am I wrong?
If I am wrong, could you please refer me to a source in which it is explained.

[FMDenaro]

I cannot see without reading the code ...
but the general idea is:

continuity: Div V = 0

momentum: dV/dt + conv + grad p = diff + b

Therefore, taking the divergence of the momentum, the pressure equation is

Div Grad p = source + Div b

with proper bc.s including the term b.n on the boundaries

[mb.pejvak]

Originally Posted by FMDenaro

I cannot see without reading the code ...
but the general idea is:

continuity: Div V = 0

momentum: dV/dt + conv + grad p = diff + b

Therefore, taking the divergence of the momentum, the pressure equation is

Div Grad p = source + Div b

with proper bc.s including the term b.n on the boundaries

Thank you Filippo for the point. I thought that all source terms should be included in "b" term of Rhie & Choe method in calculation Ue. but I was wrong. Strong source terms (body forces) should be treated as pressure. I mean as you mentioned it should be treated as a separate term. I found some papers in which it is explained.
I hope it will solve my problem.

[mb.pejvak]

finally, I applied source term on the code I developed based on changing in changing in face velocities resulted from MIM.
It needs to decrease underrelaxation factor (in my case about 0.1), but the results is not promising, particularly pressure. pressure on the surface oscillates, and it produces wrong velocity. the problem is more severe upstream of the point where source term applies, and it is confusing because source term should not have any effects in those places.
does anyone knows why it produces such oscillatory pressure?