forces function VS patchIntegrate
Hello,
I'm tried to calculate the Fx, Fy, Fz on a patch. My run is simpleFoam. With the forces function, during the calculation, for example my Fz was Fz = 13,57 kN (pressure + viscous). At the end of the run, I tried with patchIntegrate and I obtained F = 6,36 kN along a vector (912 778 5540). So Fz is Fz = 6,28 kN. This is strange! There is a 2.16 factor! Am I wronging something? Thank you for any help Andrea |
Maybe I`ve got the same problem.
see: http://www.cfd-online.com/Forums/ope...any-wrong.html Regards / Rickard |
Hi Rickard,
I already saw you post! My problem is that I used well forces function to calculate lift and drag coefficients in car analysis; but now with an ulma airplane I'm thinking the value I obtained is too much! Instead with patchIntegrate is too low! Therefore I know the pressure dimension in simpleFoam is p/rho. It could be related to our problem? Do you know if and where I could find a NACA profile to test? Thanks Andrea |
Hi Andrea.
"Do you know if and where I could find a NACA profile to test?" The equation for Naca 0009, is; 9/10*(5*(.29690*( (y)^.5)-.12600*y-.35160*((y)^2)+.28430*((y)^3)-.10150*((y)^4) ) ) A good site for wing profile is http://airfoils.worldofkrauss.com/foils for Naca 0009 http://airfoils.worldofkrauss.com/foils/1744 Regards / Rickard |
You can use one of my scripts if you like. see permissions in beginning of file
Itś at elliptic wing with ability to change angle of attack and twist of wing tip. You could use my wing channel. You had to put the WindChannel34.geo in the same directory as 20100520_2_Exp15.geo http://www.tooslow.net/rattus/CFD_Pi...dChannel34.geo http://www.tooslow.net/rattus/CFD_Pi...20_2_Exp15.geo You open 20100520_2_Exp15.geo in gmsh and do a 2D mesh then you import the saved mesh in enGrid to make a complete OpenFOAM case Good luck Regards / Rickard |
Hi see my new post on http://www.cfd-online.com/Forums/ope...any-wrong.html
I came up with a factor 2,43 Regards Rickard |
I've been having problems with an ahmed body simulation. i ran the case in Komega, Kepsilo, KomegaSST and SpalartAllmaras and in all cases i get a Cd thats somewhere between 2 and 3 times what it should be.
Maybe these problems are related.. |
I ran simultaion in both steadystate and Transcient simulation for an AUV with the different basic turbulence model. After lots of simulation to find the better scheme i reduce to 200 or 300 % of error for the Cd to 25%. But it's still high. The pressure distribution around the hull seems very good as the other variables.
Do you think that the libForces.so library may have errors ? |
Hello
Was anyone able to resolve this issue. I recently posted some other strange force calculation results in another post http://www.cfd-online.com/Forums/ope...tml#post320039 where my force calculations differ by orders of magnitude depending on whether I use a mesh generated by blockMesh or snappyHexMesh. Any insight would be greatly appreciated. Caleb |
I had similar problems but finally fixed them by choosing better div() schemes:
{Note: This was for primarily pressure drag, for viscous drag i had a lot of trouble and could never really get a good result} [check this thread, post #6: http://www.cfd-online.com/Forums/ope...fficients.html ] i switched my div(phi,u) scheme from "upwind" to "Gauss linearUpwindV cellMDLimited Gauss linear 1" and it worked. Reduced my drag error from about 120% to 10%. In that thread you can also see the dependency between the result and the mesh quality. |
Thanks for the quick reply nicolarre, but even after making your suggested change, I had already been using "Gauss linear corrected" for that term, my results are the same. I fear that something much stranger is occuring because as can be seen from the report produced by checkMesh for the mesh produced by snappyHexMesh
Code:
/*---------------------------------------------------------------------------*\ Code:
/*---------------------------------------------------------------------------*\ Caleb |
Could you attach/upload the blockMeshDict and the snappyHexMeshDict?
|
Hello nicolarre,
Here are the snappyHexMesh case files, the original 3D and the extruded 2D: http://dl.dropbox.com/u/21019547/sna..._circle.tar.gz http://dl.dropbox.com/u/21019547/sna..._circle.tar.gz I just used the half cylinder case from the basic/potential tutorial but the full case I used is here: http://dl.dropbox.com/u/21019547/hal...ockMesh.tar.gz while the full case for the snappyHexMesh is here: http://dl.dropbox.com/u/21019547/ful...HexMesh.tar.gz The snappyHexMeshDict is reproduced below Code:
/*--------------------------------*- C++ -*----------------------------------*\ Code:
/*--------------------------------*- C++ -*----------------------------------*\ Caleb |
Im' fairly new to openFoam myself, so i may or may not be of much help,
Right off the top of my head (because im not at a computer with access to openFoam) you might want to look at the aspect ratio of your blockmesh. If i recall correctly (i think its in the user manual, somewhere) snappyHexMesh dissaproves of blockmeshes with aspect ratio that aren't close to 1. The closer the blockmesh is to aspect ratio 1, the better the snappy mesh result. Speaking of aspect ratio (and taking into account i didn't get the chance to check your case yet) i see you are adding boundary layers to your snappyhexmesh. Those will usually increase your aspect ratio since they are slim. You may also want to look into your case's y+. See if its near your model's requirement. Depending on the type of simulation, viscous friction may or may not be an important factor. For my ahmed's body case, where viscous drag was low compared to preasurre drag i didn't really had much y+ restrictions. On some Airfoil cases i ran however, i had to make sure my y+ was in the right range. |
All times are GMT -4. The time now is 03:29. |