Following the previous problems I have reported with pressure over the interface of two instances I have reduced the case to a simple channel flow.
I have investigated time-scheme and time-stepping as well as mesh refinement but get no satisfactory results.
I have also looked at setting the location of the reference pressure which makes no difference either. I am thus thinking I have my case set up incorrectly.
Please could somebody look at my case files to see if I am just doing something wrong before I start to investigate a new method.
Thanks a lot,
James
Pressure drop in coupled channel flow case
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
James McNaughton
Pressure drop in coupled channel flow case
- Attachments
-
- SRC-2.tar
- (130 KiB) Downloaded 271 times
-
- SRC-1.tar
- (130 KiB) Downloaded 259 times
-
- runcase.txt
- (14.01 KiB) Downloaded 248 times
-
James McNaughton
Re: Pressure drop in coupled channel flow case
The meshes
- Attachments
-
- right_v2_2.neu
- (366.32 KiB) Downloaded 246 times
-
- left_v2_2.neu
- (366.32 KiB) Downloaded 260 times
-
Alexandre Douce
Re: Pressure drop in coupled channel flow case
Ongoing investigation. Thank you for your patience...
-
David Monfort
Re: Pressure drop in coupled channel flow case
Hello all,
Just writing here to let people know that the problem is solved AFAIK... (James correct me if I'm wrong).
The issue was because of assumptions done when coupling two domains that weren't met in James "too simple" test-case. A change in a parameter can fix the issue.
David
Just writing here to let people know that the problem is solved AFAIK... (James correct me if I'm wrong).
The issue was because of assumptions done when coupling two domains that weren't met in James "too simple" test-case. A change in a parameter can fix the issue.
David
-
James McNaughton
Re: Pressure drop in coupled channel flow case
This is correct. By simplifying a problem to test the coupling I came across issues that would not exist in the original problem :(
Unless a rotational vector (omegax,y,z != 0) is specified CS assumes coupled meshes will not be coplanar and so the global variable ifaccp = 0; in such a case the pressure on an interface is handled as a Neumann condition. By manually changing ifaccp = 1 the pressure is handled as Dirichlet and all variables are continuous over the interface.
To change ifaccp eidt the usipsu subroutine of usini1, add the following lines where appropriate:
include "cplsat.h"
ifaccp = 1
There are, however, still the issues described above with using the Neumann approach, but a Dirichlet approach is what I'd intended to use (and the Neumann approach may be intended for other uses, see https://code-saturne.info/products/code-saturne/forums/general-usage/26440831#89801714 ) and so for my case I now have a working simulation to properly test :)
Thanks,
James
Unless a rotational vector (omegax,y,z != 0) is specified CS assumes coupled meshes will not be coplanar and so the global variable ifaccp = 0; in such a case the pressure on an interface is handled as a Neumann condition. By manually changing ifaccp = 1 the pressure is handled as Dirichlet and all variables are continuous over the interface.
To change ifaccp eidt the usipsu subroutine of usini1, add the following lines where appropriate:
include "cplsat.h"
ifaccp = 1
There are, however, still the issues described above with using the Neumann approach, but a Dirichlet approach is what I'd intended to use (and the Neumann approach may be intended for other uses, see https://code-saturne.info/products/code-saturne/forums/general-usage/26440831#89801714 ) and so for my case I now have a working simulation to properly test :)
Thanks,
James