2.5D simulation in CFX-pre

yvonne · November 8, 2011, 10:47

Hi All, Im carrying out 2.5D simulation of centrifugal pump.The following is the procedure Ive used to make the geometry

As my geometry consists of 3 domains: Inlet(stationary), Impeller(rotating) and Outlet(stationary), I created the 3domains separately; i.e. 3 different ICEM files.
I followed the general 2.5D procedure: surface mesh followed by extrusion in the z-direction, separately naming the lateral faces which will be later initialized as symmetry.
Exported the 3 mesh files separately in .cfx5 format
Imported the 3 mesh files into CFX-pre
Created interfaces in CFX-pre and thus joined the 3 domains.
This worked for me and CFX simulated my 2.5D model.

Now, the problem Im facing is as follows:
There is a problem in the way CFX is calculating inlet and outlet areas. It is calculating the areas exactly an order of magnitude less than actual because of which the velocities calculated are an order of magnitude more; this results in a highly unrealistic value of pressure being calculated. Is there a way to rectify this problem? I checked out if I could write a CEL, but seems, CEL is just for post processing and for extracting calculated quantities.

Can anybody throw some light on this?

ghorrocks · November 8, 2011, 17:45

You have either a geometry/mesh problem which is making the faces too big or small, or you are scaling them in CFX-Pre.

You should always check the scale of the model is correct in CFX-Pre before proceeding.

yvonne · November 9, 2011, 05:01

Thanks for you reply. The point to be noted is, as its a 2.5D model, Im having curves instead of surfaces, so in essence CFX instead of calculating surface areas is calculating curve length (I suppose) multiplied by the one layer thickness in z-direction. Im attaching the 2D geometry. Please note that instead of outlet face Im having a 'line' and instead of inlet face Im having a 'curve'

ghorrocks · November 9, 2011, 06:18

Well, that will be your problem then. The flow rate is calculated over the one element thickness which is likely to be tiny. And I am guessing you assumed that would match the full geometry? So you need to convert the flow rates into flow per unit length.

yvonne · November 10, 2011, 02:01

Ok Thanks Glenn, youve got me thinking.. Ill think on those lines. I was considering the full flowrate

ghorrocks · November 10, 2011, 05:34

Many others have made that mistake. You are not the first.

When CFX finally introduce a proper 2D solver the problem will be fixed.

November 8, 2011, 10:47	2.5D simulation in CFX-pre	#1
yvonne New Member @p N Join Date: Jan 2010 Location: United States Posts: 27 Rep Power: 16	Hi All, Im carrying out 2.5D simulation of centrifugal pump.The following is the procedure Ive used to make the geometry As my geometry consists of 3 domains: Inlet(stationary), Impeller(rotating) and Outlet(stationary), I created the 3domains separately; i.e. 3 different ICEM files. I followed the general 2.5D procedure: surface mesh followed by extrusion in the z-direction, separately naming the lateral faces which will be later initialized as symmetry. Exported the 3 mesh files separately in .cfx5 format Imported the 3 mesh files into CFX-pre Created interfaces in CFX-pre and thus joined the 3 domains. This worked for me and CFX simulated my 2.5D model. Now, the problem Im facing is as follows: There is a problem in the way CFX is calculating inlet and outlet areas. It is calculating the areas exactly an order of magnitude less than actual because of which the velocities calculated are an order of magnitude more; this results in a highly unrealistic value of pressure being calculated. Is there a way to rectify this problem? I checked out if I could write a CEL, but seems, CEL is just for post processing and for extracting calculated quantities. Can anybody throw some light on this?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Setting Pressure Boundary Conditions in ANSYS CFX Pre	saisanthoshm88	CFX	21	February 22, 2017 17:50
Under-relaxation for steady state simulation in CFX	Chander	CFX	7	May 1, 2014 13:44
nucleate boiling simulation in CFX	Anil	CFX	3	August 25, 2010 15:18
Initial Values for Trasient simulation with CFX	shaban	CFX	1	April 30, 2010 07:25
CFX simulation file bank	eslam	CFX	2	June 15, 2007 08:46

November 8, 2011, 17:45		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,852 Rep Power: 144	You have either a geometry/mesh problem which is making the faces too big or small, or you are scaling them in CFX-Pre. You should always check the scale of the model is correct in CFX-Pre before proceeding.

November 9, 2011, 06:18		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,852 Rep Power: 144	Well, that will be your problem then. The flow rate is calculated over the one element thickness which is likely to be tiny. And I am guessing you assumed that would match the full geometry? So you need to convert the flow rates into flow per unit length.

November 10, 2011, 02:01		#5
yvonne New Member @p N Join Date: Jan 2010 Location: United States Posts: 27 Rep Power: 16	Ok Thanks Glenn, youve got me thinking.. Ill think on those lines. I was considering the full flowrate

November 10, 2011, 05:34		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,852 Rep Power: 144	Many others have made that mistake. You are not the first. When CFX finally introduce a proper 2D solver the problem will be fixed.