Simulation RANS Explode

fredicenci · May 21, 2017, 12:40

Hello all, I am trying to perform an URANS simulation (using PISO and k-omega SST) around a finite 3D cylinder and I would really appreciate any help.
Well, first of all I did a 2D simulation and worked pretty well, however when I go to the 3D domain the simulation blows, see the link with the files.

https://www.dropbox.com/…/134nlp…/AACQxNHwPovtqSx6tUYt6ylna…

I am pretty sure it is the non orthogonal faces in the end of the cylinder (checkMesh), but I can’t get rid of them ( If you take a look at the Cd in the top it goes way to high ) .. I tried to add correctors and change the schemes but it didn’t help…
Do you guys have any advice how to correct this faces or play with the correctors to get it right?
Thanks a lot!

Joshua14 · May 23, 2017, 10:04

fredicenci,

Have you tried running a lower timeStep yet?

Joshua

fredicenci · May 23, 2017, 10:52

yeh, I've tried that and it doesn't help (i decreased by half). I have done the same simulation with a coarser mesh (with no non-orthogonal faces) and worked, however i couldn't get the vortex structures I was expecting, I guess it was because the mesh was way too coarse.

Thanks for your reply =)

piu58 · May 23, 2017, 14:16

> (i decreased by half).

That may not be enough.

Start with the coarse mesh which works. The double the nodes and halve the time step. If that diverges halve the time step again until it runs.

It recommend the Courant control which some solvers offer. Even here, the usable Co may be very small for a given problem. It may be that you reach a value pretty below 1, may be even below 0.1.

The transformation of the mesh is often a very good choice. If you have (nearly) the same Co number in every part of the mesh the size of time step or usable Co number gets maximal for a given mesh resolution. If the relation of maximal to mean Co number exceeds 100 you should take this into account for sure.

fredicenci · May 23, 2017, 14:29

okay, i will try that.. I was thinking that the problem was with the mesh because the viscous layer is the same ( for the coarse mesh that worked and this one), therefore the problem could not be the Co because my Max Co would be the same on the cells near to the cylinder ... I guess this is not true...

I will reply with the results of decreasing the time-step ASAP.

Thanks for you reply! =}

May 21, 2017, 12:40	Simulation RANS Explode	#1
fredicenci Member Fredi Cenci Join Date: Dec 2016 Posts: 38 Rep Power: 9	Hello all, I am trying to perform an URANS simulation (using PISO and k-omega SST) around a finite 3D cylinder and I would really appreciate any help. Well, first of all I did a 2D simulation and worked pretty well, however when I go to the 3D domain the simulation blows, see the link with the files. https://www.dropbox.com/…/134nlp…/AACQxNHwPovtqSx6tUYt6ylna… I am pretty sure it is the non orthogonal faces in the end of the cylinder (checkMesh), but I can’t get rid of them ( If you take a look at the Cd in the top it goes way to high ) .. I tried to add correctors and change the schemes but it didn’t help… Do you guys have any advice how to correct this faces or play with the correctors to get it right? Thanks a lot!

May 23, 2017, 14:16		#4
piu58 Senior Member Uwe Pilz Join Date: Feb 2017 Location: Leipzig, Germany Posts: 744 Rep Power: 15	> (i decreased by half). That may not be enough. Start with the coarse mesh which works. The double the nodes and halve the time step. If that diverges halve the time step again until it runs. It recommend the Courant control which some solvers offer. Even here, the usable Co may be very small for a given problem. It may be that you reach a value pretty below 1, may be even below 0.1. The transformation of the mesh is often a very good choice. If you have (nearly) the same Co number in every part of the mesh the size of time step or usable Co number gets maximal for a given mesh resolution. If the relation of maximal to mean Co number exceeds 100 you should take this into account for sure. __________________ Uwe Pilz -- Die der Hauptbewegung überlagerte Schwankungsbewegung ist in ihren Einzelheiten so hoffnungslos kompliziert, daß ihre theoretische Berechnung aussichtslos erscheint. (Hermann Schlichting, 1950)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
setting up a simulation with multiple interactions	phandy	OpenFOAM Running, Solving & CFD	1	October 6, 2014 04:16
How is ANSYS CFX is related to RANS based Simulation	Ramcad	ANSYS	0	July 17, 2014 12:40
RANS simulation of Caradonna blade model	deeps	SU2	4	July 10, 2014 09:07
Discrete event simulation vs. fixed time (discrete time simulation)	jenjon	FLUENT	0	March 10, 2013 07:58
Quasi-2D mesh with gridgen for 2D RANS simulation	L1011	OpenFOAM	0	February 26, 2011 13:53

May 23, 2017, 10:04		#2
Joshua14 Member Joshua Join Date: Dec 2016 Location: St. Louis, Missouri Posts: 91 Rep Power: 9	fredicenci, Have you tried running a lower timeStep yet? Joshua

May 23, 2017, 10:52		#3
fredicenci Member Fredi Cenci Join Date: Dec 2016 Posts: 38 Rep Power: 9	yeh, I've tried that and it doesn't help (i decreased by half). I have done the same simulation with a coarser mesh (with no non-orthogonal faces) and worked, however i couldn't get the vortex structures I was expecting, I guess it was because the mesh was way too coarse. Thanks for your reply =)

May 23, 2017, 14:29		#5
fredicenci Member Fredi Cenci Join Date: Dec 2016 Posts: 38 Rep Power: 9	okay, i will try that.. I was thinking that the problem was with the mesh because the viscous layer is the same ( for the coarse mesh that worked and this one), therefore the problem could not be the Co because my Max Co would be the same on the cells near to the cylinder ... I guess this is not true... I will reply with the results of decreasing the time-step ASAP. Thanks for you reply! =}