Favorize

Hamed&honey · October 8, 2012, 05:23

Hi everybody,
See the attached file, I have a sloped solid body, when I favorize the model with inefficient mesh size. I see the stepped slope and it’s not good, I know about FAVOR action, but how I can improve my model (except increasing mesh).
Thanks

mderosca · October 8, 2012, 08:06

Hello,

As I know is just visualization. Favor algorithm creates a plane in each cell to determinate the surfaces, but when makes the render don't shown the planes and apears littles steps.

So run the model and you will see no problems.

Hamed&honey · October 8, 2012, 11:16

dear mderosca
I think your answer is not correct, after I run the simulation still have problems, please check it and tell me if i wrong.
so many thanks

robert_cfd · October 9, 2012, 04:29

Hi,

For such a simple case you can make the geometry horizontal and change the gravity instead. Just a workaround for simple geometries, otherwise gridrefinement and a simulation restart.

Hamed&honey · October 9, 2012, 06:04

dear robert_cfd,
thank you for answer
But my model is so complicated (Ogee spillway with chute)
even use 2 million volume still I have step in my model that is bad influence on my real model.

robert_cfd · October 9, 2012, 08:12

Ok, I see. Is the spillway straight? Just the spillway, not the whole powerplant? When I look at your attached picture, you might also use symmetry or even just one cell in width (quasi 2D). Then you can use more cells without loosing time.

Hamed&honey · October 9, 2012, 18:37

I have 3D model and you can see the whole 3d solid in attached file.
with 2 million mesh still I have problem,

robert_cfd · October 10, 2012, 09:16

For me it looks like you could test
-how it works with one cell in y direction
-or rotate the geometry around y axis till the flat main part is horizontal, using the angle for calculation of the x and z component of the new gravity field. Then you only have to take care of the in- and outflow area with a finer grid.

Davahue · October 17, 2012, 05:49

Dear Hammed,
in my experience, you would be impressed with results. Maybe you don't expect such good ones.. I would launch some 2D simulations only to prove it.
But maybe you have tried yet. If you are asking for something like unstructures body fitted meshes you won't find it. Just give it a chance

Quote:

Originally Posted by Hamed&honey

I have 3D model and you can see the whole 3d solid in attached file.
with 2 million mesh still I have problem,

Hamed&honey · October 17, 2012, 06:39

dear Davahue
I have investigating cavitation on spillway chute, and we know that the rough surface can be increase cavitation index, I have straight slope in real, but stepped slope in model. is this ok? I think its not going to accurate result. am I wrong?
Thanks

Davahue · October 17, 2012, 11:47

You are wright,
but I suppose you are using a RANS model, so the way it faces the boundary layer has nothing to do with what you are thinking (from a classical hydraulical point of view). I would recommend you to read about wall functions in David Wilcox's (or others) or the FlowScience Technical Note about Roughness. The aproximation all RANS turbulence models do about the surface roughness is very important, if you take it into account, you won't be so demanding on surface resolution. I mean, for a numerical model of N-S equations to reflect exact gradients nearby the surface would need thousands of cells in the boundary layer. If you have not that resolution, that artificial roughness created by FAVOR, in my opinion, is not afecting definitively the main flow because the model doesn't catch it.

I'm not an expert, but i'm sure Michael Barkhudarov or many other experts in this forum could tell you if i'm wright or wrong or how to face your problem better. Anyway, try it and compare the results with experimental data.

Hope it helps,
D.

QUOTE=Hamed&honey;387049]dear Davahue
I have investigating cavitation on spillway chute, and we know that the rough surface can be increase cavitation index, I have straight slope in real, but stepped slope in model. is this ok? I think its not going to accurate result. am I wrong?
Thanks[/QUOTE]

Hamed&honey · October 17, 2012, 12:48

dear Davahue
I think you are an expert

in fact, basic problem is about velocity,
Those steps are very effective on velocity
cause to reduction.
I compare with experimental model.

velocity at the jump location
Exp=25 m/s
CFD=20 m/s

October 8, 2012, 08:06	Favor	#2
mderosca New Member mderosca Join Date: Jul 2012 Posts: 5 Rep Power: 14	Hello, As I know is just visualization. Favor algorithm creates a plane in each cell to determinate the surfaces, but when makes the render don't shown the planes and apears littles steps. So run the model and you will see no problems.

October 8, 2012, 11:16		#3
Hamed&honey New Member Ham Ta Join Date: May 2010 Posts: 18 Rep Power: 16	dear mderosca I think your answer is not correct, after I run the simulation still have problems, please check it and tell me if i wrong. so many thanks

October 9, 2012, 04:29		#4
robert_cfd New Member Join Date: Oct 2012 Posts: 3 Rep Power: 14	Hi, For such a simple case you can make the geometry horizontal and change the gravity instead. Just a workaround for simple geometries, otherwise gridrefinement and a simulation restart.

October 9, 2012, 06:04		#5
Hamed&honey New Member Ham Ta Join Date: May 2010 Posts: 18 Rep Power: 16	dear robert_cfd, thank you for answer But my model is so complicated (Ogee spillway with chute) even use 2 million volume still I have step in my model that is bad influence on my real model.

October 9, 2012, 08:12		#6
robert_cfd New Member Join Date: Oct 2012 Posts: 3 Rep Power: 14	Ok, I see. Is the spillway straight? Just the spillway, not the whole powerplant? When I look at your attached picture, you might also use symmetry or even just one cell in width (quasi 2D). Then you can use more cells without loosing time.

October 10, 2012, 09:16		#8
robert_cfd New Member Join Date: Oct 2012 Posts: 3 Rep Power: 14	For me it looks like you could test -how it works with one cell in y direction -or rotate the geometry around y axis till the flat main part is horizontal, using the angle for calculation of the x and z component of the new gravity field. Then you only have to take care of the in- and outflow area with a finer grid.

October 17, 2012, 06:39		#10
Hamed&honey New Member Ham Ta Join Date: May 2010 Posts: 18 Rep Power: 16	dear Davahue I have investigating cavitation on spillway chute, and we know that the rough surface can be increase cavitation index, I have straight slope in real, but stepped slope in model. is this ok? I think its not going to accurate result. am I wrong? Thanks

October 17, 2012, 11:47		#11
Davahue Member Daniel Join Date: Sep 2010 Location: the Netherlands Posts: 44 Rep Power: 16	You are wright, but I suppose you are using a RANS model, so the way it faces the boundary layer has nothing to do with what you are thinking (from a classical hydraulical point of view). I would recommend you to read about wall functions in David Wilcox's (or others) or the FlowScience Technical Note about Roughness. The aproximation all RANS turbulence models do about the surface roughness is very important, if you take it into account, you won't be so demanding on surface resolution. I mean, for a numerical model of N-S equations to reflect exact gradients nearby the surface would need thousands of cells in the boundary layer. If you have not that resolution, that artificial roughness created by FAVOR, in my opinion, is not afecting definitively the main flow because the model doesn't catch it. I'm not an expert, but i'm sure Michael Barkhudarov or many other experts in this forum could tell you if i'm wright or wrong or how to face your problem better. Anyway, try it and compare the results with experimental data. Hope it helps, D. QUOTE=Hamed&honey;387049]dear Davahue I have investigating cavitation on spillway chute, and we know that the rough surface can be increase cavitation index, I have straight slope in real, but stepped slope in model. is this ok? I think its not going to accurate result. am I wrong? Thanks[/QUOTE]

October 17, 2012, 12:48		#12
Hamed&honey New Member Ham Ta Join Date: May 2010 Posts: 18 Rep Power: 16	dear Davahue I think you are an expert in fact, basic problem is about velocity, Those steps are very effective on velocity cause to reduction. I compare with experimental model. velocity at the jump location Exp=25 m/s CFD=20 m/s