Regarding BiGlobal stability in Nektar++
Hi, I am trying to understand the session file "PPF_R15000_3D_VCS.xml." I have two confusion: 1) Since modeType is HalfMode than why "HomModesZ" is set to 2. I think this should be 1. 2) Since the base flow is 2D2C, then is it necessary to include variable "w" in the base flow? In general, can we use a base flow generated from DNS which only have variables (uvp) in the session file having FIELDS = (uvwp)? I would be highly thankful for the comments on this issue. With regards Abhishek Kumar --------------------------------------------------------------------------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ---------------------------------------------------------------------------------------
Hi Abhishek, Have you tried with setting this value to 1, and did the solver work and returned the same result? 1. Anyway, as far as I understand what is in SolverUtils/EquationSystem.cpp the value of HomModesZ is overwritten on initialization: 2 for SingleMode and 1 for HalfMode - so the value in the session file would seem redundant (Solver starts with this value removed). 2. Yes, AFAIK you need to have w in the BaseFlow section. It does not need to come from a file, you could do sth. like this: <F VAR="u,v,p" FILE="bflow.fld" /> <E VAR="w" VALUE="0" /> So you can use a 2D solution with Fourier expansion. Best, Stan Gepner On czw, 2018-10-25 at 22:31 +0100, Abhishek Kumar wrote:
Hi,
I am trying to understand the session file "PPF_R15000_3D_VCS.xml." I have two confusion: 1) Since modeType is HalfMode than why "HomModesZ" is set to 2. I think this should be 1.
2) Since the base flow is 2D2C, then is it necessary to include variable "w" in the base flow?
In general, can we use a base flow generated from DNS which only have variables (uvp) in the session file having FIELDS = (uvwp)?
I would be highly thankful for the comments on this issue.
With regards Abhishek Kumar
------------------------------------------------------------------- -------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom
Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ------------------------------------------------------------------- -------------------- _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Hi Stanislaw, Thanks for the prompt reply. 1) Yes, you are correct. There is no effect of HomModesZ. It gives the same result with HomModesZ=1. 2) For the base flow, it works with <F VAR="u,v,p" FILE="bflow.fld" />. It doesn't ask for w. But the problem comes when we add a scalar variable (for a problem like Rayleigh-Benard). In this case, I use <F VAR="u,v,c,p" FILE="base.fld" />, and it gives the following error: Fatal : Level 0 assertion violation For half mode, expect 2D2C base flow. I attached the session file. Can you comment on this issue? With regards Abhishek On Fri, Oct 26, 2018 at 8:29 AM Stanisław Gepner <sgepner@meil.pw.edu.pl> wrote:
Hi Abhishek,
Have you tried with setting this value to 1, and did the solver work and returned the same result?
1. Anyway, as far as I understand what is in SolverUtils/EquationSystem.cpp the value of HomModesZ is overwritten on initialization: 2 for SingleMode and 1 for HalfMode - so the value in the session file would seem redundant (Solver starts with this value removed).
2. Yes, AFAIK you need to have w in the BaseFlow section. It does not need to come from a file, you could do sth. like this:
<F VAR="u,v,p" FILE="bflow.fld" /> <E VAR="w" VALUE="0" />
So you can use a 2D solution with Fourier expansion.
Best, Stan Gepner
On czw, 2018-10-25 at 22:31 +0100, Abhishek Kumar wrote:
Hi,
I am trying to understand the session file "PPF_R15000_3D_VCS.xml." I have two confusion: 1) Since modeType is HalfMode than why "HomModesZ" is set to 2. I think this should be 1.
2) Since the base flow is 2D2C, then is it necessary to include variable "w" in the base flow?
In general, can we use a base flow generated from DNS which only have variables (uvp) in the session file having FIELDS = (uvwp)?
I would be highly thankful for the comments on this issue.
With regards Abhishek Kumar
------------------------------------------------------------------- -------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom
Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ------------------------------------------------------------------- -------------------- _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- With regards Abhishek Kumar --------------------------------------------------------------------------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ---------------------------------------------------------------------------------------
Hi Abhishek, You can try the following approach: 1. Have a 2D solution with u,v,c,p. 2. Prepare a session file with NumSteps = 0, and a homogeneous expansion in the z direction. 3. Add w to the expansion, variables boundary, etc. In initial conditions read u, v, c, p from file and set w to zero. 4. Run it, and you get a solution with u,v,w,c,p. Note: I am not sure you can successfully do this with HalfMode. SingleMode will work. Note 2: I am not sure this will be usable for the stability solver as baseflow (although I see no reason why?). I have not used the stability solver with an "active" scalar. I just played with the Rayleigh-Benard problem after David Moxley activated the ability to formulate BodyForce using the scalar value, about a year ago. Best, Stan On pią, 2018-10-26 at 11:31 +0100, Abhishek Kumar wrote:
Hi Stanislaw,
Thanks for the prompt reply.
1) Yes, you are correct. There is no effect of HomModesZ. It gives the same result with HomModesZ=1. 2) For the base flow, it works with <F VAR="u,v,p" FILE="bflow.fld" />. It doesn't ask for w.
But the problem comes when we add a scalar variable (for a problem like Rayleigh-Benard). In this case, I use <F VAR="u,v,c,p" FILE="base.fld" />, and it gives the following error:
Fatal : Level 0 assertion violation For half mode, expect 2D2C base flow.
I attached the session file. Can you comment on this issue?
With regards Abhishek
On Fri, Oct 26, 2018 at 8:29 AM Stanisław Gepner < sgepner@meil.pw.edu.pl> wrote:
Hi Abhishek,
Have you tried with setting this value to 1, and did the solver work and returned the same result?
1. Anyway, as far as I understand what is in SolverUtils/EquationSystem.cpp the value of HomModesZ is overwritten on initialization: 2 for SingleMode and 1 for HalfMode - so the value in the session file would seem redundant (Solver starts with this value removed).
2. Yes, AFAIK you need to have w in the BaseFlow section. It does not need to come from a file, you could do sth. like this:
<F VAR="u,v,p" FILE="bflow.fld" /> <E VAR="w" VALUE="0" />
So you can use a 2D solution with Fourier expansion.
Best, Stan Gepner
On czw, 2018-10-25 at 22:31 +0100, Abhishek Kumar wrote:
Hi,
I am trying to understand the session file "PPF_R15000_3D_VCS.xml." I have two confusion: 1) Since modeType is HalfMode than why "HomModesZ" is set to 2. I think this should be 1.
2) Since the base flow is 2D2C, then is it necessary to include variable "w" in the base flow?
In general, can we use a base flow generated from DNS which only have variables (uvp) in the session file having FIELDS = (uvwp)?
I would be highly thankful for the comments on this issue.
With regards Abhishek Kumar
---------------------------------------------------------------
-------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom
Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ---------------------------------------------------------------
-------------------- _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Hi Stanisław, Thanks for your reply. I will try the first approach. I remembered that David activated the BodyForce using the scalar value. Solver works perfectly for 2D and 3D DNS. It also works for the stability for 2D base flow with 2D perturbation. Problem is with the HalfMode. Again thanks for your suggestions. With regards Abhishek On Fri, Oct 26, 2018 at 12:49 PM Stanisław Gepner <sgepner@meil.pw.edu.pl> wrote:
Hi Abhishek,
You can try the following approach: 1. Have a 2D solution with u,v,c,p. 2. Prepare a session file with NumSteps = 0, and a homogeneous expansion in the z direction. 3. Add w to the expansion, variables boundary, etc. In initial conditions read u, v, c, p from file and set w to zero. 4. Run it, and you get a solution with u,v,w,c,p.
Note: I am not sure you can successfully do this with HalfMode. SingleMode will work.
Note 2: I am not sure this will be usable for the stability solver as baseflow (although I see no reason why?). I have not used the stability solver with an "active" scalar. I just played with the Rayleigh-Benard problem after David Moxley activated the ability to formulate BodyForce using the scalar value, about a year ago.
Best, Stan
On pią, 2018-10-26 at 11:31 +0100, Abhishek Kumar wrote:
Hi Stanislaw,
Thanks for the prompt reply.
1) Yes, you are correct. There is no effect of HomModesZ. It gives the same result with HomModesZ=1. 2) For the base flow, it works with <F VAR="u,v,p" FILE="bflow.fld" />. It doesn't ask for w.
But the problem comes when we add a scalar variable (for a problem like Rayleigh-Benard). In this case, I use <F VAR="u,v,c,p" FILE="base.fld" />, and it gives the following error:
Fatal : Level 0 assertion violation For half mode, expect 2D2C base flow.
I attached the session file. Can you comment on this issue?
With regards Abhishek
On Fri, Oct 26, 2018 at 8:29 AM Stanisław Gepner < sgepner@meil.pw.edu.pl> wrote:
Hi Abhishek,
Have you tried with setting this value to 1, and did the solver work and returned the same result?
1. Anyway, as far as I understand what is in SolverUtils/EquationSystem.cpp the value of HomModesZ is overwritten on initialization: 2 for SingleMode and 1 for HalfMode - so the value in the session file would seem redundant (Solver starts with this value removed).
2. Yes, AFAIK you need to have w in the BaseFlow section. It does not need to come from a file, you could do sth. like this:
<F VAR="u,v,p" FILE="bflow.fld" /> <E VAR="w" VALUE="0" />
So you can use a 2D solution with Fourier expansion.
Best, Stan Gepner
On czw, 2018-10-25 at 22:31 +0100, Abhishek Kumar wrote:
Hi,
I am trying to understand the session file "PPF_R15000_3D_VCS.xml." I have two confusion: 1) Since modeType is HalfMode than why "HomModesZ" is set to 2. I think this should be 1.
2) Since the base flow is 2D2C, then is it necessary to include variable "w" in the base flow?
In general, can we use a base flow generated from DNS which only have variables (uvp) in the session file having FIELDS = (uvwp)?
I would be highly thankful for the comments on this issue.
With regards Abhishek Kumar
---------------------------------------------------------------
-------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom
Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ---------------------------------------------------------------
-------------------- _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- With regards Abhishek Kumar --------------------------------------------------------------------------------------- Abhishek Kumar Research Associate Applied Mathematics Research Centre Coventry University, Coventry CV1 5FB The United Kingdom Email # ac7600@coventry.ac.uk, abhishek.kir@gmail.com ---------------------------------------------------------------------------------------
participants (2)
-
Abhishek Kumar
-
Stanisław Gepner