Periodic boundary condition vs FNMB

Cappytbg · March 15, 2018, 05:39

Hello!

Can someone resolve my doubts about difference between periodic and FNMB condition?

My issue: I am trying to perform analysis with two sectors with unstructured grid in Numeca/Fine Open.
I am connecting those sectors with FNMB condition (periodic one), but then Numeca Open sees it as a "wall" between fluids. Wall on which I am supposed to set a boundary condition.

1. I am curious how Numeca treats usual periodic boundary condition PER - it doesnt appear in boundary conditions as a solid. Can you tell me if it is "automaticly" adiabatic or what?
I couldnt find it anywhere in User/Theory guide.

2. Is there any other way to connect two fluid interfaces?

Thanks for every help!

Kacper

DarylMusashi · March 15, 2018, 07:23

Dear Kacper,

1
It is a normal behaviour that the FNMBs appear in the Boundary Condition -> Solids Tab. The situation is that the defined FNMB consists of two sides. At the overlapping part of both sides the fluid can flow through it.

2.
The PER boundaries are typical periodic conditions. Very colloquially spoken: If something enters the periodic boundary it appears on the other related periodic boundary. Is this the intention of your question?

3.
Another possibility to connect two fluid regions are rotor-stator interfaces. Maybe you could share a sketch of your machine?

Kind regards
Holger

Cappytbg · March 15, 2018, 09:12

Dear Holger,

Thank you for your reply!
Now when I look from perspective you gave, it is obvious that PER boundary is not a wall... I was treating it like a wall.

I need to connect two passages because my computer is not able to make grid for two passages simultaneously, I thought I will create one mesh and then import it, but then I cannot use PER boundary.

Maybe there is some way to connect those two imported meshes?

Are FNMB much worse than PER? Do you know?

sketch of the problem: https://imgur.com/a/imfUC

DarylMusashi · March 15, 2018, 15:22

Dear Kacper,

I am not sure if I understood you correctly. In your sketch I see 3 rotor domains, not 2. I see two green ones (low and middle) and a blue one (top). I think you want to connect the blue one with the two green ones, correct?

May I ask why you mesh 2 stator and 3 rotor passage? The only reason which comes to my mind is you want to run an unsteady simulation, which requires an equal pitch of stator row and rotor row.

However, the interpolation error of a FNMB generally depends strongly on the cell sizes on both sides. If the cell sizes are nearly equal the interpolation error should be acceptable.

Kind regards,
Holger

Cappytbg · March 16, 2018, 04:14

Dear Holger,

Sorry it is only sketch, as I said - it is 2 rotor domains and 1 stator. Each color is for single domain and black lines are boundaries.
I am connecting rotor and stator in usual way - by rotor/stator interface, but my issue is to connect also two rotor domains.
Reason is that I am a student. And I am researching, if there is a difference between this way and the other, but you are right I am also going to do an unsteady simulation.

Thank you, for your advice with this interpolation, it is also opening new thinking.

The last question - do you happen to know if there is any way to "merge" two grids in Hexpress? Like I am doing - importing meshes separately and then merge them?

DarylMusashi · March 16, 2018, 06:41

Dear Kacper,

ok I see, thanks to clearify this point. You said your computer hasn't enough performance to mesh both passages at once. I suppose the mesh will be too large and you run out of memory (RAM)?
But secondly you are going to perform unsteady simulations. I assume you won't do this on your low performance PC. Maybe it could be possible to do the meshing on the machine on which you perform the (unsteady) simulations?

"Merging" like you intent it is not possible in HEXPRESS, you can only attach both meshes to each other with FNMBs.

May I ask which meshing software you use? It seems you mesh both rotor domains separately in another program and import them in HEXPRESS.

Looking forward to hearing from you,
Holger

Cappytbg · March 16, 2018, 07:52

Dear Holger,

Yes, with this meshing, RAM is the problem. But I think for simulations mainly CPUs are significant, is it right?
Yes, I also have not found "merging" option in Hexpress. I think FNMB will be some option for me.

I am using Hexpress for meshing all domains and then I am also able to import "Hexpress" mesh into Hexpress.

Big thank you for your help Holger!

Kind regards,
Kacper

March 15, 2018, 05:39	Periodic boundary condition vs FNMB	#1
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Hello! Can someone resolve my doubts about difference between periodic and FNMB condition? My issue: I am trying to perform analysis with two sectors with unstructured grid in Numeca/Fine Open. I am connecting those sectors with FNMB condition (periodic one), but then Numeca Open sees it as a "wall" between fluids. Wall on which I am supposed to set a boundary condition. 1. I am curious how Numeca treats usual periodic boundary condition PER - it doesnt appear in boundary conditions as a solid. Can you tell me if it is "automaticly" adiabatic or what? I couldnt find it anywhere in User/Theory guide. 2. Is there any other way to connect two fluid interfaces? Thanks for every help! Kacper

March 15, 2018, 07:23		#2
DarylMusashi Senior Member Holger Dietrich Join Date: Apr 2011 Location: Germany Posts: 174 Rep Power: 14	Dear Kacper, 1 It is a normal behaviour that the FNMBs appear in the Boundary Condition -> Solids Tab. The situation is that the defined FNMB consists of two sides. At the overlapping part of both sides the fluid can flow through it. 2. The PER boundaries are typical periodic conditions. Very colloquially spoken: If something enters the periodic boundary it appears on the other related periodic boundary. Is this the intention of your question? 3. Another possibility to connect two fluid regions are rotor-stator interfaces. Maybe you could share a sketch of your machine? Kind regards Holger hakuhiko likes this.

March 15, 2018, 15:22		#4
DarylMusashi Senior Member Holger Dietrich Join Date: Apr 2011 Location: Germany Posts: 174 Rep Power: 14	Dear Kacper, I am not sure if I understood you correctly. In your sketch I see 3 rotor domains, not 2. I see two green ones (low and middle) and a blue one (top). I think you want to connect the blue one with the two green ones, correct? May I ask why you mesh 2 stator and 3 rotor passage? The only reason which comes to my mind is you want to run an unsteady simulation, which requires an equal pitch of stator row and rotor row. However, the interpolation error of a FNMB generally depends strongly on the cell sizes on both sides. If the cell sizes are nearly equal the interpolation error should be acceptable. Kind regards, Holger Last edited by DarylMusashi; March 16, 2018 at 06:59.

March 16, 2018, 06:41		#6
DarylMusashi Senior Member Holger Dietrich Join Date: Apr 2011 Location: Germany Posts: 174 Rep Power: 14	Dear Kacper, ok I see, thanks to clearify this point. You said your computer hasn't enough performance to mesh both passages at once. I suppose the mesh will be too large and you run out of memory (RAM)? But secondly you are going to perform unsteady simulations. I assume you won't do this on your low performance PC. Maybe it could be possible to do the meshing on the machine on which you perform the (unsteady) simulations? "Merging" like you intent it is not possible in HEXPRESS, you can only attach both meshes to each other with FNMBs. May I ask which meshing software you use? It seems you mesh both rotor domains separately in another program and import them in HEXPRESS. Looking forward to hearing from you, Holger Last edited by DarylMusashi; March 16, 2018 at 08:31.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
sliding mesh problem in CFX	Saima	CFX	46	September 11, 2021 07:38
Problem in setting Boundary Condition	Madhatter92	CFX	12	January 12, 2016 04:39
Unsteady Periodic Boundary Condition	EMolina	SU2	0	July 24, 2014 16:16
About periodic boundary condition.	kohel_11	FLUENT	3	July 30, 2013 07:37
Error finding variable "THERMX"	sunilpatil	CFX	8	April 26, 2013 07:00

March 15, 2018, 09:12		#3
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Dear Holger, Thank you for your reply! Now when I look from perspective you gave, it is obvious that PER boundary is not a wall... I was treating it like a wall. I need to connect two passages because my computer is not able to make grid for two passages simultaneously, I thought I will create one mesh and then import it, but then I cannot use PER boundary. Maybe there is some way to connect those two imported meshes? Are FNMB much worse than PER? Do you know? sketch of the problem: https://imgur.com/a/imfUC

March 16, 2018, 04:14		#5
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Dear Holger, Sorry it is only sketch, as I said - it is 2 rotor domains and 1 stator. Each color is for single domain and black lines are boundaries. I am connecting rotor and stator in usual way - by rotor/stator interface, but my issue is to connect also two rotor domains. Reason is that I am a student. And I am researching, if there is a difference between this way and the other, but you are right I am also going to do an unsteady simulation. Thank you, for your advice with this interpolation, it is also opening new thinking. The last question - do you happen to know if there is any way to "merge" two grids in Hexpress? Like I am doing - importing meshes separately and then merge them?

March 16, 2018, 07:52		#7
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Dear Holger, Yes, with this meshing, RAM is the problem. But I think for simulations mainly CPUs are significant, is it right? Yes, I also have not found "merging" option in Hexpress. I think FNMB will be some option for me. I am using Hexpress for meshing all domains and then I am also able to import "Hexpress" mesh into Hexpress. Big thank you for your help Holger! Kind regards, Kacper