CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > ANSYS > CFX

cavitation bubbles

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   August 28, 2017, 09:11
Default cavitation bubbles
  #1
New Member
 
ozz
Join Date: Aug 2017
Posts: 6
Rep Power: 8
ozanx is on a distinguished road
Hello dear forum users;

Im trying to simulate the bubble occuration with a transient simulation on cfx for a sonotrode which have 20kHz frequency and 40um displacement. My problem is whenever I simulate it I got irrelevant bubble points ( bubbles with isovolume above 0.15 vapor volume fraction). I dont know where is the problem. Do I need extra formulas inserted for this type of analysis?

Thanks in advance
ozanx is offline   Reply With Quote

Old   August 28, 2017, 18:57
Default
  #2
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,749
Rep Power: 143
ghorrocks is just really niceghorrocks is just really niceghorrocks is just really niceghorrocks is just really nice
Have you looked at the FAQ on accuracy? https://www.cfd-online.com/Wiki/Ansy..._inaccurate.3F

Have you checked your mesh, time step and convergence are adequate to resolve the features you are looking for?
ghorrocks is offline   Reply With Quote

Old   August 29, 2017, 05:24
Default
  #3
New Member
 
ozz
Join Date: Aug 2017
Posts: 6
Rep Power: 8
ozanx is on a distinguished road
Hello
Yes Ive checked these things you mentioned. My problem is the cavitation bubbles (determined by isovolume which states the volumes containing more than 0.05 vapor volume fraction) are only occuring on the surface which is vibrating, but normally with the real experiments there are also vapor bubbles away from the surface. In my analysis vapors dont leave the surface.
ozanx is offline   Reply With Quote

Old   August 29, 2017, 05:46
Default
  #4
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,749
Rep Power: 143
ghorrocks is just really niceghorrocks is just really niceghorrocks is just really niceghorrocks is just really nice
You say you have checked mesh, time step and convergence but 95% of the time the problem is those factors - can you explain what you did to check them?

How many cycles of the motion have you run this for?
ghorrocks is offline   Reply With Quote

Old   August 29, 2017, 08:57
Default
  #5
New Member
 
ozz
Join Date: Aug 2017
Posts: 6
Rep Power: 8
ozanx is on a distinguished road
I checked the mesh with the mesh quality tool. I simply try to make a finer mesh around critical areas using sizing and other mesh tools, my goal was to improve mesh quality around these areas trying different methods.

My converge criteria was 1e-5 as suggested in manuals of this page.

Although Im not sure about the time step. I made the analysis with 20 timesteps for every period. ( 1 sinusoidal wave )
ozanx is offline   Reply With Quote

Old   August 29, 2017, 19:07
Default
  #6
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,749
Rep Power: 143
ghorrocks is just really niceghorrocks is just really niceghorrocks is just really niceghorrocks is just really nice
For the mesh you need to consider both the quality and the resolution. Try a finer mesh.

20 time steps per period is unlikely to be enough.

For all these tunable parameters (mesh density, convergence tolerance, time step size) you should do a sensitivity analysis to set them correctly. If you guess you will invariably get it wrong and get unreliable results - like what you are reporting.
ghorrocks is offline   Reply With Quote

Old   August 30, 2017, 03:29
Default
  #7
New Member
 
ozz
Join Date: Aug 2017
Posts: 6
Rep Power: 8
ozanx is on a distinguished road
I left the analysis run for 15 cycles last night with a finer mesh, but again against the real experiment the vapor dont leave the surface. Do I need another setup for this
ozanx is offline   Reply With Quote

Old   August 30, 2017, 06:21
Default
  #8
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,749
Rep Power: 143
ghorrocks is just really niceghorrocks is just really niceghorrocks is just really niceghorrocks is just really nice
Let me explain sensitivity studies.....

You do a simulation and get a result. Then you change the parameter of interest, let's say that is mesh size. You must do a change big enough to be significant - for mesh size that would be halving the edge length (which will result in around 8 times as many nodes, so a much bigger mesh). So make a new mesh with half the element edge length and do the simulation again. Compare the results to the first simulation.

If the results are the same within a tolerance you are happy with then you have found an adequate mesh size and can proceed. If the results are different then you must refine the mesh again, halving the element edge length again and continue until you do obtain results close enough for your purposes.

For convergence tolerance each step should be a factor of 10, for time step each step should be a factor of 2. But a short cut is to use adaptive time stepping homing in on 3-5 coeff loops per iteration. That will find a good time step by itself.

If you start by finding a good mesh size with a guessed convergence tolerance and time step size, then you need to check the convergence tolerance and time step size. This may mean you original mesh is no longer adequate - so you need to repeat the mesh sensitivity study to check the mesh is OK.

You can see this can be an iterative process, and can be time consuming especially for complex models. But unless you do it you have no idea if your answers are accurate or not - and if you just guessed the parameters they are almost certainly totally wrong.
ghorrocks is offline   Reply With Quote

Old   August 31, 2017, 03:15
Default
  #9
New Member
 
ozz
Join Date: Aug 2017
Posts: 6
Rep Power: 8
ozanx is on a distinguished road
Is solving the problem multiple times take too much time? Since only one takes nearly 3-5 hours to complete.
ozanx is offline   Reply With Quote

Old   August 31, 2017, 03:27
Default
  #10
Super Moderator
 
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,749
Rep Power: 143
ghorrocks is just really niceghorrocks is just really niceghorrocks is just really niceghorrocks is just really nice
That's why people do CFD on supercomputers. It's up to you - either do it properly and it will take a bit of time and effort, or cut corners and get rubbish results.
ghorrocks is offline   Reply With Quote

Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
why doesn't ansys allow a negative pressure when cavitation is included vysje CFX 16 December 11, 2016 17:18
cavitation problem for a vibrating plate in water ndabir FLUENT 1 May 9, 2016 20:08
fetal overflow in user defined cavitation model unclewallcn CFX 15 January 20, 2016 22:17
Cavitation in Pure Ethanol Pugnax CFX 4 June 18, 2015 20:19
Combining Cavitation and Thermal Effects? akash_max CFX 4 January 19, 2012 14:09


All times are GMT -4. The time now is 09:24.