Which model can I use instead of RNG k-E and get good pressures?

danijela.djuric.mijovic · June 16, 2014, 09:56

Hallo,

Which turbulent model should I use instead of RNG k-E and get good pressures on the building facades? I had good results and good convergence while windward wall was parallel to Inlet plane and when the building is turned for 15 degrees.
But when the angle is 30 or 45 degrees RNG has really bad convergence and not stable solution.

Thanks,
Danijela

ghorrocks · June 16, 2014, 22:09

You change turbulence models because of accuracy issues, not numerical stability issues.

I would recommend you use the SST turbulence model unless you have a good reason to do otherwise.

If you are having troubles getting convergence look at this FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

danijela.djuric.mijovic · June 17, 2014, 10:45

Thanks Glenn
for suggestion. I used also SST besides K-Omega for some previous models but usually for initializaion file. And SST gives better pressures on the building facades then k-Omega. The RNG has the best one.
the thing is that the approach I had for cases when building is 0 and 15 degree turned can not be applied for 30 and 45 degrees. With the same approach (probably mesh is the main thing) I can get only 10e-3 or 10e-4 (RMS) much worse then for first two cases.

ghorrocks · June 17, 2014, 19:58

Have you read the FAQ? It tells you everything you need to know about this issue.

danijela.djuric.mijovic · June 20, 2014, 11:04

Yes I did

not once

but most of advices consider transient analysis and I have steady state. But this was interesting ' If these parameters are not changing to an accuracy tolerance suitable for your simulation then your simulation is almost certainly OK'. I was monitoring the velocity and pressure and they oscilate. So I stopped analysis in the maximum and minimum and compared...

the difference is not big up to 2.5%, in one point it is 5 and on leeward wall is 16%. well I have to check if I stopped it in the right moment

ghorrocks · June 21, 2014, 06:04

You did not read it very carefully then - it is written for when a steady state simulation does not converge. That is the title. The transient simulation is simply one of the options to obtain convergence if it is particularly stubborn - but it still applies to a steady state simulation.

The best thing to do to monitor convergence like this is to have monitor points with CEL expressions which calculate your parameter of interest. Then you can watch the convergence of them as the simulation progresses and you do not need to stop it to see how it is going.

June 16, 2014, 09:56	Which model can I use instead of RNG k-E and get good pressures?	#1
danijela.djuric.mijovic New Member Danijela Djuric Mijovic Join Date: Feb 2014 Posts: 8 Rep Power: 12	Hallo, Which turbulent model should I use instead of RNG k-E and get good pressures on the building facades? I had good results and good convergence while windward wall was parallel to Inlet plane and when the building is turned for 15 degrees. But when the angle is 30 or 45 degrees RNG has really bad convergence and not stable solution. Thanks, Danijela

June 16, 2014, 22:09		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	You change turbulence models because of accuracy issues, not numerical stability issues. I would recommend you use the SST turbulence model unless you have a good reason to do otherwise. If you are having troubles getting convergence look at this FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

June 17, 2014, 10:45		#3
danijela.djuric.mijovic New Member Danijela Djuric Mijovic Join Date: Feb 2014 Posts: 8 Rep Power: 12	Thanks Glenn for suggestion. I used also SST besides K-Omega for some previous models but usually for initializaion file. And SST gives better pressures on the building facades then k-Omega. The RNG has the best one. the thing is that the approach I had for cases when building is 0 and 15 degree turned can not be applied for 30 and 45 degrees. With the same approach (probably mesh is the main thing) I can get only 10e-3 or 10e-4 (RMS) much worse then for first two cases.

June 17, 2014, 19:58		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	Have you read the FAQ? It tells you everything you need to know about this issue.

June 20, 2014, 11:04		#5
danijela.djuric.mijovic New Member Danijela Djuric Mijovic Join Date: Feb 2014 Posts: 8 Rep Power: 12	Yes I did not once but most of advices consider transient analysis and I have steady state. But this was interesting ' If these parameters are not changing to an accuracy tolerance suitable for your simulation then your simulation is almost certainly OK'. I was monitoring the velocity and pressure and they oscilate. So I stopped analysis in the maximum and minimum and compared... the difference is not big up to 2.5%, in one point it is 5 and on leeward wall is 16%. well I have to check if I stopped it in the right moment

June 21, 2014, 06:04		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	You did not read it very carefully then - it is written for when a steady state simulation does not converge. That is the title. The transient simulation is simply one of the options to obtain convergence if it is particularly stubborn - but it still applies to a steady state simulation. The best thing to do to monitor convergence like this is to have monitor points with CEL expressions which calculate your parameter of interest. Then you can watch the convergence of them as the simulation progresses and you do not need to stop it to see how it is going.