
[Sponsors] 
December 14, 2011, 14:48 
Problem with Sampling plane

#1 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
Hi,
I observe that for a given plane, mass flow through that plane calculated using massFlow function is different (sometimes very sigificantly) if the plane is of Slice or of Sample type. I have taken care to have same location and bounds for the plane in both the types. Still massflow for a Slice type of plane is correct as per my simulation and the same for Sample type is wrong. The reason I why I want to use Sample plane is because I want to calculate mass flow through planar, nonoverlapping stripes of equal size made normal to flow direction so as to have mass flow distribution along coordinate normal to flow direction. Refer attached figure which shows the stripes along coordinate y while the flow is along xdirection. Now the massflow calculated through these stripes is wrong because it does not addup to the total mass flow over the entire y extent at that location. I cannot use Slice plane to make these stripes it does not result in stripes of equal y sizes as my mesh is nonuniform in ydirection. How can I solve this problem ? Thanks for your inputs 

December 14, 2011, 14:53 

#2 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
Just to claify, flow is along xdirection and I have made equal stripes along ydirection using Sample plane i.e. normal to the flow. Total mass flow through these stripes is 1.233e3 kg/s while the correct value is 6.24e3 kg/s


December 14, 2011, 17:44 

#3 
Senior Member
Edmund Singer P.E.
Join Date: Aug 2010
Location: Minneapolis, MN
Posts: 512
Rep Power: 12 
Try increasing the number of points in the sample plane. It is probably too coarse.


December 14, 2011, 19:13 

#4 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
I have used the maximum allowed sample points on each stripe.


February 16, 2012, 14:15 

#5 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
Hello everyone,
Sorry to revive this thread after 1 month or so. I have still not solved this problem and Ansys support has not been of much help on this issue. I just want to find the mass flow distribution along the y direction in the figure attached above. Any ideas on how to get that will be extremely helpful Best regards, 

February 16, 2012, 18:02 

#6 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 13,209
Rep Power: 103 
I suspect Edmund is on the right track, the sampling poitn density is not high enough. If you have already maxxed it out then you might have to start looking for nasty workarounds  maybe chop the surface into lots of little ones, and add up the contribution from each surface. If the answer is heading towards the correct value you know you are on the right track.


February 16, 2012, 20:55 

#7 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
Hello Glen,
Yes, I am already maxed out on the number of points. By chopping the surface, do u mean increasing the number of sampling planes? One more thing. I also tried with the same number of sampling planes but on the surface marked as inlet (refer the figure above) instead of the position shown in figure (i.e. the sampling planes were xy planes on the inlet surface rather than the yz planes as in figure). If I sum the mass flow rate through all the planes in the second case (i.e. using xy sampling planes on inlet surface), I get total fow rate close to the actual value. But as I already stated previously, I get a very different and wrong value in case of the yz planes shown in the figure. Why should this happen? The total flow rate should come out to be same in both the cases. 

February 17, 2012, 05:22 

#8 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 13,209
Rep Power: 103 
Yes, I mean multiple sampling planes.
The point Edmund and I are trying to make is that the sampling approach is not fine enough to accurately resolve the flow, but the slice option uses the elemental data and is accurate. How you slice and dice is up to you but the point Edmund and I are trying to make remains the same  insufficient density of sampling points. 

February 20, 2012, 08:22 

#9  
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
Well, I noticed something new.
Please refer to the new attached figure As I show in this figure, fluid is supplied to the inlet plane in the older figure from a fluid duct. The fluid duct is a fluid domain and the blue colored portion is porous domain. Now when I stated the following in my previous post: Quote:
Now I found out that if I take the following sum on yz planes: sum over all i ( mass flow rate through ith plane * masflowAve(volume porosity) on ith plane) then this sum comes close (within 2%) to the actual mass flow rate ! In essence while calculating the mass flow rate over each yz sampling plane, I had to weight the mass flow rate with the corresponding average volume porosity at that location ! I did not have to use this weighting when the sampling planes were located inside fluid domain (and in fluid domain, value of volume porosity would anyway be 1 everywhere if one were to use the weighting). Remember that I cannot use a series of slice planes instead of sampling planes as that leads to planes of unequal size in ydirection due to mesh nonuniformity. I also tested the above conclusion over a yz plane across the entire y extent of the porous domain. at the same x location as before. I observe that: a) If that plane is of slice type, then massflow@ that plane gives the correct mass flow rate value b) If that plane if of sample type, then massflow@plane * massflowAve(Volume porosity) @plane comes close to the actual massflow rate. So the question is that why should there be a need to weigh with porosity while calculating massflow rate through a sampling plane (and not while using a slice plane) located inside a porous medium? 

February 20, 2012, 17:51 

#10 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 13,209
Rep Power: 103 
I see, things are clearer now. Try areaAve(massflow*Volume Porosity)@plane or something like that, that might be a little more accurate again.


February 20, 2012, 17:57 

#11 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 8 
I have already tried the areaAve and that gives a value similar to massflowAve i.e. within 2% of the correct value.
But the question remains that while finding mass flow through a plane inside porous domain, why does slice plane give the correct value directly while one needs to multiply with averaged porosity to get the same if one uses a sampling plane instead of the slice plane. 

Tags 
cfxpost 
Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
sampling a plane  Gearb0x  OpenFOAM PostProcessing  5  December 17, 2012 10:49 
Meshing a Mixing Plane in ICEM  Will Anderson  ANSYS Meshing & Geometry  2  November 9, 2010 10:16 
Velocity profiles problem behind the elbow (3D problem)  kabat73  FLUENT  8  May 9, 2010 04:26 
Import problem  ARC  Open Source Meshers: Gmsh, Netgen, CGNS, ...  0  February 27, 2010 11:56 
Adiabatic and Rotating wall (Convection problem)  ParodDav  CFX  5  April 29, 2007 19:13 