-
Notifications
You must be signed in to change notification settings - Fork 71
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
NoSlipWall in pelelm and pelec #502
Comments
Please post your PeleC input file. Thanks! |
I have tried many cases, but this problem can't be solved. |
Thanks for providing the input file. The first thing I see is that you have |
Oops, switch to pelec.diffuse_vel = 1 :) |
Thanks a million. |
What flow quantity are you plotting here? What does your input file look like? Do you have this in the input file?
|
Temperature. I want to set isothermal boundaries for the tube wall and EB, but it doesn't work very well |
@baperry2 you've done heat conduction with EB in the past to verify the capability. Do you see anything wrong with this input file? |
It looks like you want gradients at the cartesian face at the bottom. If you set it as no-slip wall, your if(idir==1) conditions from bcnormal will be ignored. By default the "noslipwall" is adiabatic. if you want to set it as an isothermal wall, use "Hard" there. pelec.lo_bc = "Hard" "Hard" |
@czc-zju the inputs file you posted should be sufficient for setting the isothermal condition at the EB surface. Let me know if you ware still having trouble at the EB surface after implementing these changes. You can also see regtests EB-C4-5 for examples with an isothermal EB boundary. The changes @hsitaram recommends should fix the boundary condition at the lower boundary. |
great! It would be better to do an odd reflection, ie s_ext[UMX]=-s_int[UMX] etc.. for momentum and similarly for temperature as s_ext[UTEMP]=2*500-s_int[UTEMP] |
Do you mean set like this: s_ext[URHO] = s_int[URHO]; |
yes, looks good! |
Fantastic!!! |
looks like you set pelec.diffuse_vel=0, it should be 1 if you want viscous
effects
…On Sun, May 22, 2022 at 7:03 PM czc-zju ***@***.***> wrote:
example_inp.txt
<https://github.com/AMReX-Combustion/PeleC/files/8750622/example_inp.txt>
—
Reply to this email directly, view it on GitHub
<#502 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABYOQMYF4XI7UQCMNTLJK43VLLKPDANCNFSM5WSXBKUQ>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
@hsitaram just clarifying things here Aren't UTEMP and UEINT supposedly "derived" variables as per the docs? So shouldn't we be setting the Isothermal BC in a conservative sense based on UEDEN rather than UTEMP? I tried a very simple case involving a rectangular channel with the top and bottom walls at 2500K and air inside the channel at zero velocity with an initial temperature of 300K like so: Zero gradient for UEDEN and UEINT and the dirchlet for UTEMP didn't seem to cause any rise in temperature of the air inside. However, setting the same BC in a conservative sense on UEDEN like the following does seem to work:
|
Hi Surya,
Thank you for bringing this up.
UTEMP is actually not in the list of derived variables. Setting UEDEN like
the way you did can change the hyperbolic flux at the wall, which is why I
suggested the UTEMP fix.
if you change UEDEN, you change the pressure at the wall, resulting in a
non-zero pressure gradient (dpdn), which is not correct.
Therefore, changing only UTEMP would impose just a thermal diffusive flux
and no changes to hyperbolics.
That said, as of now, changing UTEMP in bcnormal is not changing anything,
as the wall is behaving like it is adiabatic.
Although, commenting out this line -
https://github.com/AMReX-Combustion/PeleC/blob/ff51a0c37af5fc74e0baad3ae06991f93a7e6704/Source/Utilities.H#L231
should fix the thermal diffusion flux without any issues. Let me discuss
with my colleagues here before deeming this change safe. I can see the
correct behavior, see below:
![Screen Shot 2022-10-13 at 10 47 11 PM](https://user-images.githubusercontent.com/7399475/195764859-670704d8-d7b7-4f72-8af1-c3420a20c227.png)
|
@hsitaram will this change work? |
Hi Surya,
I had a discussion with my colleague Marc H and the hack (removing the
REY2T call) I mentioned earlier should be ok.
I checked if there were any changes to regression tests with this change.
They were negligible, at least for the PMF case.
I strongly suggest not implementing the isothermal wall bc through UEDEN as
you will affect the hyperbolic flux.
We also discussed if we need to add a new BC for isothermal wall along with
slip/noSlip BCs,
which would require some effort on our side to perform checks while
assigning ghost cell quantities. will keep you posted on this.
It will be good to hear from you again if it works out for you.
Thanks,
Hari
…On Mon, Oct 17, 2022 at 12:02 PM RSuryaNarayan ***@***.***> wrote:
@hsitaram <https://github.com/hsitaram> will this change work?
—
Reply to this email directly, view it on GitHub
<#502 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABYOQM7ZT3VYONNV25E3LV3WDWICZANCNFSM5WSXBKUQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Hello Hari, The case I am running is similar to what I showed in the graphic in my previous comment except that the wall is now NoSlip for which I plan to simply reflect the velocities. Thanks once again! |
Hello Hari |
Is this only for evaluating the gradient? That is, will the T that comes from that formula be used to evaluate any EOS or transport coefficient calls? |
@drummerdoc we use that to set |
that ghost T does go into EOS calls. So we need to be careful here.
s_int[UTEMP] may be the safest option.
…On Fri, Oct 28, 2022 at 12:11 PM Marc Day ***@***.***> wrote:
Is this only for evaluating the gradient? That is, will the T that comes
from that formula be used to evaluate any EOS or transport coefficient
calls?
—
Reply to this email directly, view it on GitHub
<#502 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABYOQM62NTIOE7GCUEOFL73WFQJL3ANCNFSM5WSXBKUQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Wouldn't that set it back to adiabatic wall? For now something that seems to work is just T_wall |
my bad. use s_ext[UTEMP] to be T_wall.
…On Fri, Oct 28, 2022 at 1:57 PM RSuryaNarayan ***@***.***> wrote:
Would that set it back to adiabatic wall?
—
Reply to this email directly, view it on GitHub
<#502 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABYOQM23D3GJ3ARK2VYKNKDWFQV33ANCNFSM5WSXBKUQ>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Thanks! |
The following two cases are calculated by PeleLM and PeleC. In the results of PeleLM, we can see clear boundary layers, the velocity is very small, but the result of PeleC has no boundary layer, and the velocity on the vertical line is equal, why?
The text was updated successfully, but these errors were encountered: