Problems scaling down the mesh of a stirred tank (GGI, icoDyMFoam)

Tobb · August 24, 2010, 08:27

Hello everybody!

For this is my first post in this forum, I hope I opened this thread in the right place.

I' ve got a problem simulating the flow in a stirred tank using icoDyMFoam and GGI.
The mesh consists of one cylinder (height 20cm, diameter 15cm), the shaft (also a cylinder, diameter 5mm) and the impeller with for blades (diameter 2,5cm).
I created the mesh in Gambit, but with the dimensions in meters, I intended to scale them afterwards using "transformpoints".

When I ran my case for the first time, I forgot to scale my mesh so I used the values in meters and not in centimeters. However the results where pretty fine, the velocity field and distribution of pressure looked the way I expected. (rpm = 60 1/minute)

But when I scaled my case to the desired dimensions and changed the rotational frequency to 30 rpm (as my tutor recommended), my simulation got really bad. The velocity field got kind of asymmetric, my maximum pressure became even negativ!

I "played around" trying different values for the rotational frequency. I managed to get good results with a case being just ten times smaller then my original mesh (as I said I need a case 100 times smaller because of converting my mesh from m to cm) using rpm = 600 1/minute. This is equivalent to the same peripheral velocity of the impeller as in the first case. Yet I didn't consider this to be the right way because I shouldn't need to achieve the same peripheral velocity but rather the same reynolds number, it is to say, the same outer conditions for the flow. Especially in order to stay in laminar region as icoDyMFoam ist a solver for incompressible and laminar flow as i know.
For stirred tanks I found the formula
Re = (f * d^2)/nu
But this would mean my frequency would grow quadratically. Concering my case I would have to increase my frequency by a factor of 10000, to me this seems to be nonsense.

I have the impression that I am overlooking a fundamental fault and could ruin my reputation before I really have one.

I suppose there are some rules for "down-scaling" that I am not aware of. So I would be glad if someone could slightly point out to me, where I have to search my mistake.
Thank you in advance,

Tobb

stainboy · January 31, 2013, 12:12

Hello Tobb,

have you found solution for your problem? Since OpenFOAM respects units I'm wondering if when I would like to have my solution scaled to cm, is it necessary just to scale viscosity to cm^2/s, pressure to s^2/cm^2 and velocity to cm/s.

Kuba.

Alucard · February 7, 2013, 14:06

If your mesh is in meters just scale the velocity (or the viscosity) in order to have the right Re number you desire.

If you have problems increasing your Re number start with lower values and increment it until you start seeing "weird things" happening in your system and analyse the mesh resolution and/or the advective scheme used.

August 24, 2010, 08:27	Problems scaling down the mesh of a stirred tank (GGI, icoDyMFoam)	#1
Tobb New Member Join Date: Aug 2010 Posts: 6 Rep Power: 16	Hello everybody! For this is my first post in this forum, I hope I opened this thread in the right place. I' ve got a problem simulating the flow in a stirred tank using icoDyMFoam and GGI. The mesh consists of one cylinder (height 20cm, diameter 15cm), the shaft (also a cylinder, diameter 5mm) and the impeller with for blades (diameter 2,5cm). I created the mesh in Gambit, but with the dimensions in meters, I intended to scale them afterwards using "transformpoints". When I ran my case for the first time, I forgot to scale my mesh so I used the values in meters and not in centimeters. However the results where pretty fine, the velocity field and distribution of pressure looked the way I expected. (rpm = 60 1/minute) But when I scaled my case to the desired dimensions and changed the rotational frequency to 30 rpm (as my tutor recommended), my simulation got really bad. The velocity field got kind of asymmetric, my maximum pressure became even negativ! I "played around" trying different values for the rotational frequency. I managed to get good results with a case being just ten times smaller then my original mesh (as I said I need a case 100 times smaller because of converting my mesh from m to cm) using rpm = 600 1/minute. This is equivalent to the same peripheral velocity of the impeller as in the first case. Yet I didn't consider this to be the right way because I shouldn't need to achieve the same peripheral velocity but rather the same reynolds number, it is to say, the same outer conditions for the flow. Especially in order to stay in laminar region as icoDyMFoam ist a solver for incompressible and laminar flow as i know. For stirred tanks I found the formula Re = (f * d^2)/nu But this would mean my frequency would grow quadratically. Concering my case I would have to increase my frequency by a factor of 10000, to me this seems to be nonsense. I have the impression that I am overlooking a fundamental fault and could ruin my reputation before I really have one. I suppose there are some rules for "down-scaling" that I am not aware of. So I would be glad if someone could slightly point out to me, where I have to search my mistake. Thank you in advance, Tobb

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[snappyHexMesh] SnappyHexMesh for internal Flow	vishwa	OpenFOAM Meshing & Mesh Conversion	24	June 27, 2016 09:54
[ICEM] Unstructure Meshing Around Imported Plot3D Structured Mesh ICEM	kawamatt2	ANSYS Meshing & Geometry	17	December 20, 2011 12:45
Mesh problems for a stirred vessel with baffles!	UY	CFX	0	September 14, 2005 11:19
How to control Minximum mesh space?	hung	FLUENT	7	April 18, 2005 10:38
unstructured vs. structured grids	Frank Muldoon	Main CFD Forum	1	January 5, 1999 11:09

January 31, 2013, 12:12		#2
stainboy New Member Jakub Pola Join Date: Feb 2011 Posts: 22 Rep Power: 15	Hello Tobb, have you found solution for your problem? Since OpenFOAM respects units I'm wondering if when I would like to have my solution scaled to cm, is it necessary just to scale viscosity to cm^2/s, pressure to s^2/cm^2 and velocity to cm/s. Kuba.

February 7, 2013, 14:06		#3
Alucard Member valerio Join Date: Jun 2009 Location: Nancy Posts: 30 Rep Power: 17	If your mesh is in meters just scale the velocity (or the viscosity) in order to have the right Re number you desire. If you have problems increasing your Re number start with lower values and increment it until you start seeing "weird things" happening in your system and analyse the mesh resolution and/or the advective scheme used.