CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Couette Flow (https://www.cfd-online.com/Forums/main/201135-couette-flow.html)

FMDenaro May 12, 2018 13:57

Quote:

Originally Posted by IbizaFR (Post 692125)
Thank you FMDenaro.

So for all cells I check the max value of divergence and if not satisfied according to the maximum permitted value then I continue to new iteration.
Is it right?




do not forget that you have a system of equations therefore you have to satisfy continuity and momentum equation simultaneously

IbizaFR May 13, 2018 04:29

Thank you FMDenaro.
What is the checking condition for the momentum?

FMDenaro May 13, 2018 04:42

The residuals of the two momentum equations

IbizaFR May 13, 2018 04:46

Do you mean the absolute change in u,v from one itetation to the next one?

FMDenaro May 13, 2018 04:47

Quote:

Originally Posted by IbizaFR (Post 692154)
Do you mean the absolute change in u,v from one itetation to the next one?


I mean the residuals of the equations. The difference between two iterations is not the residual.

IbizaFR May 13, 2018 04:53

Thank you.
The check has to be done in the provisional velocities or to the corrected in SIMPLE?

IbizaFR May 13, 2018 05:38

Something else.
Do I have to use periodic boundary conditions mandatorily?

FMDenaro May 13, 2018 05:59

Quote:

Originally Posted by IbizaFR (Post 692157)
Something else.
Do I have to use periodic boundary conditions mandatorily?


It would be better

IbizaFR May 13, 2018 06:17

Can I avoid them with other way?

FMDenaro May 13, 2018 06:25

Quote:

Originally Posted by IbizaFR (Post 692163)
Can I avoid them with other way?




The theoretical solution requires two plates estending infinitely in both directions.

Using fictious inlet/outlet BC.s requires that you prescribe the exact solution at inflow and a developed condition at outlet. I never tried such BC.s but I suppose they will drive to the exact solution at the steady state.

IbizaFR May 13, 2018 07:02

So at the inflow and outflow I can give linear distribution of u and p=0 at both ends. Is it right?

IbizaFR May 13, 2018 13:09

Thank you FMDenaro,

How can I implement periodic boundary conditions in this problem? Can you suggest me a book to study?

IbizaFR May 14, 2018 04:02

Thank you FMDenaro,

Because I have not dealt with periodic boundary conditions before can you suggest me a book to study about this matter?

FMDenaro May 14, 2018 04:20

Quote:

Originally Posted by IbizaFR (Post 692229)
Thank you FMDenaro,

Because I have not dealt with periodic boundary conditions before can you suggest me a book to study about this matter?


You cannot find a complete book about the periodic boundary conditions... it is a numerical issue that is simply implemented in a code by using a link between boundary nodes.
I give you an example in 1D for a periodicity over a lenght L. You define the nodes 1,2...N in such a way that x(i)=(i-1)*h+Xs. This way x(N)=Xs+L.
Now, you have that nodes 1 and N are linked by periodicity. You compute the equation either in node 1 or N and prescribe the computed value at the node you do not compute.

IbizaFR May 14, 2018 06:35

Thank you,

Let me ask you:
What is Xs?
Do I have to alter all the momentum and pressure equations in SIMPLE at two ends or only the u,v,p at these ends?

FMDenaro May 14, 2018 06:39

Quote:

Originally Posted by IbizaFR (Post 692241)
Thank you,

Let me ask you:
What is Xs?
Do I have to alter all the momentum and pressure equations in SIMPLE at two ends or only the u,v,p at these ends?




Xs is only the general value of the position of the left point, it could be also 0. The periodicity involves all the variables.

IbizaFR May 14, 2018 06:41

Quote:

Originally Posted by FMDenaro (Post 692243)
Xs is only the general value of the position of the left point, it could be also 0. The periodicity involves all the variables.

I cann't understand if I write x(N)=Xs+L with Xs=0 so X(N)=L, how the code will understand that the start and finish node are the same?

FMDenaro May 14, 2018 06:47

Quote:

Originally Posted by IbizaFR (Post 692244)
I cann't understand if I write x(N)=Xs+L with Xs=0 so X(N)=L, how the code will understand that the start and finish node are the same?


Solve all the equations at i=1 and set the obtained values at i=N

IbizaFR May 14, 2018 06:53

So all the loops will not calculate anything at the i=N and set the calculated values at i=1 to i=N. Is it right this?

Thank you

FMDenaro May 14, 2018 06:58

Quote:

Originally Posted by IbizaFR (Post 692246)
So all the loops will not calculate anything at the i=N and set the calculated values at i=1 to i=N. Is it right this?

Thank you




yes, correct


All times are GMT -4. The time now is 09:17.