Change blend factor for specific domains

p.mueller · January 24, 2019, 10:49

Hello everybody,

I'm simulating a CHT-Modell from a turbocharger. Now I would like to change the advection scheme (the specific blend factor) for the Enclosure. So that my Simulation uses for the Enclosure and ambient Air an Upwind (the specific blend factor = 0) Discretization and for the inner Fluid a High Res Discretization (the specific blend factor = 0.75).

Is this possible and do you have an idea how can I write in in an expression or something else.
Thank you very much.

Regards
Paul

Opaque · January 24, 2019, 11:21

May I ask why you would like to have "errors" in a section of the model, and accurate in the other?

p.mueller · January 24, 2019, 14:49

With that consideration, I would like to do an error analysis for the enclosure. I would like to compare a second order (High Res Discretization) with a first order procedure like Upwind. In the 2nd order method, the error minimizes quadratically when halving the mesh size. If I use a first order method with the same mesh, the error reduces linearly. So it halves the error by halving the mesh size.

ghorrocks · January 24, 2019, 17:22

You could try using a CEL variable for the blend factor. I suspect it won't let you. Normal practice is to apply the same advection scheme to the entire model. I have never heard of somebody wanting to change it in different regions of the model.

Also consider schemes like High Resolution. This uses a second order scheme where ever it can and where it estimates that there will be too much instability it rolls back to first order. It does this automatically with no need for users to specify where. I would recommend this scheme over what you propose.

p.mueller · January 25, 2019, 15:27

That make sense. Thank you

. It's more a try for myself. So I would try to use a CEL variable. Which condition I could use in the CEL? Most times I would use the time step as a condition, but this makes no sense. Do you have maybe an idea.

ghorrocks · January 25, 2019, 19:05

As we said previously, your request to change blend factor per domain is not recommended practice. The recommended approach is to either make it constant over the entire simulation, or use a scheme like HiRes which automatically adjusts the order of the advection scheme.

But if you still want to do it, you could use the inside() CEL function to determine which domain you are in. You could also use an if() statement using some function of the geometry (eg x>10[m]).

January 24, 2019, 10:49	Change blend factor for specific domains	#1
p.mueller New Member Join Date: Aug 2018 Posts: 7 Rep Power: 8	Hello everybody, I'm simulating a CHT-Modell from a turbocharger. Now I would like to change the advection scheme (the specific blend factor) for the Enclosure. So that my Simulation uses for the Enclosure and ambient Air an Upwind (the specific blend factor = 0) Discretization and for the inner Fluid a High Res Discretization (the specific blend factor = 0.75). Is this possible and do you have an idea how can I write in in an expression or something else. Thank you very much. Regards Paul

January 24, 2019, 17:22		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	You could try using a CEL variable for the blend factor. I suspect it won't let you. Normal practice is to apply the same advection scheme to the entire model. I have never heard of somebody wanting to change it in different regions of the model. Also consider schemes like High Resolution. This uses a second order scheme where ever it can and where it estimates that there will be too much instability it rolls back to first order. It does this automatically with no need for users to specify where. I would recommend this scheme over what you propose. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

January 25, 2019, 19:05		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	As we said previously, your request to change blend factor per domain is not recommended practice. The recommended approach is to either make it constant over the entire simulation, or use a scheme like HiRes which automatically adjusts the order of the advection scheme. But if you still want to do it, you could use the inside() CEL function to determine which domain you are in. You could also use an if() statement using some function of the geometry (eg x>10[m]). __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[ANSYS Meshing] Displaying solid domains in CFD Post without meshing them.	hda	ANSYS Meshing & Geometry	5	October 24, 2016 10:26
Does not assemble the domains to make a block although it is a close volume	mollaee.saeed	Pointwise & Gridgen	5	October 8, 2015 04:46
Radiation interface	hinca	CFX	15	January 26, 2014 18:11
Porous domains in contact	0906536m	CFX	3	September 30, 2013 19:34
How to change turbulence model in InterFoam	Gildeh	OpenFOAM Running, Solving & CFD	4	March 28, 2012 13:04

January 24, 2019, 11:21		#2
Opaque Senior Member Join Date: Jun 2009 Posts: 1,873 Rep Power: 33	May I ask why you would like to have "errors" in a section of the model, and accurate in the other?

January 24, 2019, 14:49		#3
p.mueller New Member Join Date: Aug 2018 Posts: 7 Rep Power: 8	With that consideration, I would like to do an error analysis for the enclosure. I would like to compare a second order (High Res Discretization) with a first order procedure like Upwind. In the 2nd order method, the error minimizes quadratically when halving the mesh size. If I use a first order method with the same mesh, the error reduces linearly. So it halves the error by halving the mesh size.

January 25, 2019, 15:27		#5
p.mueller New Member Join Date: Aug 2018 Posts: 7 Rep Power: 8	That make sense. Thank you . It's more a try for myself. So I would try to use a CEL variable. Which condition I could use in the CEL? Most times I would use the time step as a condition, but this makes no sense. Do you have maybe an idea.