How can I detect cavitation presence while the simulation is still running?

Stabum · May 16, 2015, 05:53

Hello
I need to design the blades for an axial pump whose flow rate, head and RPM is known.
Obviously the pump should not caviate.

My question is double:
1) Let's suppose I could "probe" the static pressure on the most critical region of the pump blade and that I could monitor this variable during the simulation process. At what residual level (or at what convergence level) do the local static pressure (in case of cavitation of course) become reliable? I mean: should I wait to reach 1E-7 MAX residual or 1E-4 is enough for this kind of information?
2) The blade grid has been created with TGrid. Is there a way I could DYNAMICALLY monitor a variable (like "static pressure" in the most critical region of the blade) during the simulation, so that if I see that the value gets stable (and near to 2500 Pa) I can say: ok, there's cavitation, I need to modify the blades, let's stop the simulation!

I hope I've been "understandable". If not, then please feel free to ask and I'll try to be more clear.

As Always, thanks so much for your help!!!

ghorrocks · May 16, 2015, 18:44

1) the convergence required to achieve a reliable answer is obviously different for different applications. For some people 1% error could be all they can tolerance, but for others being off by 20% is OK. On top of that, a given level of residual does not equate to a universal accuracy number - the relationship between accuracy and residuals is not fixed.

This is a long way of saying you should determine the level of residual required for your application and your accuracy requirement. Do a sweep of convergence tolerance and monitor values of importance to you - then you can select the tolerance which gives you the accuracy you require.

2) Use a monitor point. You could put it at a defined point like you suggest, or you could just use minVal(p)@domain and it will return the minimum pressure in the entire domain.

Stabum · May 17, 2015, 09:59

Thanks so much for your super useful answer. I understand that residuals and convergence are always delicate concepts. Can you tell me something more about the monitor point? How can I locate it? It would be perfect for my case. Thanks again

ghorrocks · May 17, 2015, 19:08

Monitor points are described in the documentation and used in the tutorial examples.

Have a look on the output tab in CFX-Pre.

Stabum · May 18, 2015, 00:57

Ok, thanks. I'm checking out right now. Thank again!!!

Stabum · May 18, 2015, 18:38

I used the monitor points as a probe in given locations on the blade, as you suggested. Everything works so nice! Thanks, ghorrocks, you saved my life!

May 16, 2015, 05:53	How can I detect cavitation presence while the simulation is still running?	#1
Stabum Member Join Date: Aug 2012 Location: Italy Posts: 66 Rep Power: 13	Hello I need to design the blades for an axial pump whose flow rate, head and RPM is known. Obviously the pump should not caviate. My question is double: 1) Let's suppose I could "probe" the static pressure on the most critical region of the pump blade and that I could monitor this variable during the simulation process. At what residual level (or at what convergence level) do the local static pressure (in case of cavitation of course) become reliable? I mean: should I wait to reach 1E-7 MAX residual or 1E-4 is enough for this kind of information? 2) The blade grid has been created with TGrid. Is there a way I could DYNAMICALLY monitor a variable (like "static pressure" in the most critical region of the blade) during the simulation, so that if I see that the value gets stable (and near to 2500 Pa) I can say: ok, there's cavitation, I need to modify the blades, let's stop the simulation! I hope I've been "understandable". If not, then please feel free to ask and I'll try to be more clear. As Always, thanks so much for your help!!!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pimpleFoam deltaT increases rapidly running TJunction like simulation?	massive_turbulence	OpenFOAM Running, Solving & CFD	8	April 24, 2014 21:09
CFD simulation of centrifugal pump cavitation	billy7590	Fluent Multiphase	0	March 22, 2014 08:28
CFD simulation of centrifugal pump cavitation	billy7590	Main CFD Forum	0	March 22, 2014 02:27
Cavitation Simulation	Carolina Parrinelli	Main CFD Forum	7	September 25, 2012 00:03
Cavitation Simulation by CFD	Liu, L.	Main CFD Forum	7	November 1, 2000 21:26

May 16, 2015, 18:44		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	1) the convergence required to achieve a reliable answer is obviously different for different applications. For some people 1% error could be all they can tolerance, but for others being off by 20% is OK. On top of that, a given level of residual does not equate to a universal accuracy number - the relationship between accuracy and residuals is not fixed. This is a long way of saying you should determine the level of residual required for your application and your accuracy requirement. Do a sweep of convergence tolerance and monitor values of importance to you - then you can select the tolerance which gives you the accuracy you require. 2) Use a monitor point. You could put it at a defined point like you suggest, or you could just use minVal(p)@domain and it will return the minimum pressure in the entire domain.

May 17, 2015, 09:59		#3
Stabum Member Join Date: Aug 2012 Location: Italy Posts: 66 Rep Power: 13	Thanks so much for your super useful answer. I understand that residuals and convergence are always delicate concepts. Can you tell me something more about the monitor point? How can I locate it? It would be perfect for my case. Thanks again

May 17, 2015, 19:08		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	Monitor points are described in the documentation and used in the tutorial examples. Have a look on the output tab in CFX-Pre.

May 18, 2015, 00:57		#5
Stabum Member Join Date: Aug 2012 Location: Italy Posts: 66 Rep Power: 13	Ok, thanks. I'm checking out right now. Thank again!!!

May 18, 2015, 18:38		#6
Stabum Member Join Date: Aug 2012 Location: Italy Posts: 66 Rep Power: 13	I used the monitor points as a probe in given locations on the blade, as you suggested. Everything works so nice! Thanks, ghorrocks, you saved my life!