AnsweredAssumed Answered

Jet ski pump solver terminated

Question asked by Patrk Rećko on Apr 10, 2016
Latest reply on Apr 14, 2016 by Patrk Rećko

Like • Show 0 Likes0
Comment • 13

Hello everyone,
I am new on this forum so firstly let me introduce myself. My name is Patryk and I am a student of Mechanical engineering in Bialystok, Poland.
I'm working on my school project which is a flow in jet ski drive, an example of axial water pump. My problem is that I getting an error at about 4500 iteration (~2/4 travels): "Solver abnormally terminated! Please contact the support service", but when I'm solving the same simulation but on scaled model, two or three times smaller, there is no error, solver finishing normally.
I tried to do internal and external flows, low and high quality mesh and problem is always the same, solver terminates on true scale model, but on two or three times smaller it doesn't.

I'm working with a cavitation model turned on and rotating region.
I noticed that a while before error appears, density, velocity and pressure suddenly increase their values and sometimes I'm getting an error "Real gas parameters (pressure and/or temperature) are outside the definitional domain of substance properties T<Tmin".
Anyone could help me ?
Best regards,
Patryk

Amit Katz Apr 12, 2016 12:07 PM

Correct Answer

It seems to me from your goals chart that your critical parameters are more or less converged before the automatic refinement, is there a particular reason you need so many iterations? You could also go to "Calculation Control Options" and manually disable the refinement.

See the reply in context

No one else had this question

Outcomes

Amit Katz Apr 11, 2016 9:24 AM

This error is a catch-all error that means "I give up!" for the software. From how you describe your parameters spiking I'm guessing the solution diverges and the model becomes unsolvable. Possibly the mesh is too coarse. And if you need more than 4500 iterations to solve your model, then it's probably not set up correctly. Any chance you can upload your project?

Actions

Patrk Rećko @ Amit Katz on Apr 11, 2016 11:51 AM

Zippyshare.com - 4000RPM cavitation ON.rar
There is the model with 5000 RPM's set up on impeller (name of folder is wrong), I uploaded it with that invaid result, maybe that will help. I noticed that it doesn't depend on iterations, it always getting an error after 2.1-2.2 travels. Weird thing is that I can solve this case up to 3000RPM even on true scale model but if the value is higher - solver terminates. Finally I want to do this simulation with 8300 RPM's.
1:2 scale model with 4150 RPM and 8 points automatic mesh finish normally after 8900 iterations (internal analysis).

Actions

Amit Katz @ Patrk Rećko on Apr 11, 2016 12:29 PM

If you save your file without the CFD results then you can upload it directly to the forums. In general to help you we don't really need your results, just the setup.

Actions

Patrk Rećko @ Amit Katz on Apr 11, 2016 12:45 PM

There is only model with setup

Attachments

5000rpm.rar19.4 MB

Actions

Amit Katz @ Patrk Rećko on Apr 11, 2016 4:31 PM

The file was 130 MB and I was able to load your CFD results, so it looks like you copied everything in. For future reference you can use the "Pack and Go" feature to save an assembly into a zip file and choose whether or not to include certain things like simulation results.

Actions

David Paulson @ Patrk Rećko on Apr 11, 2016 6:38 PM

Patryk,

I was not able to open your download file, but from your experience the solver works on a 50% +/- scale model but not on a full scale model at about 5000 RPM. Possible issues:

1. If the impeller diameter or RPM is not scaled you will not be seeing the same fluid velocities? A 50 mm impeller scaled to 25 mm will produce much lower pressures and velocities at the same RPM.

2. You did not state the number of cells in your simulation and the comparison between full scale and 1/2 scale. I have had similar issues with simulations with 4-5 million cells, until I increased the DRAM in my computer. More DRAM seemed to fix my problem.

3. 4500 iterations seems like a lot. Solver is struggling for convergence. Did you, or can you, display the solver solution graph?

Actions

Patrk Rećko @ David Paulson on Apr 12, 2016 4:13 AM

There is less iterations because of reducing the size of computional domain and poor mesh. If there is more iterations the section on graph between 200 and ~450 is longer and almost straight. Solver always terminates a while after this little white dot appears on "iterations axis" and before it appears solver actually achieving all of the goals at 100%.

Actions

Amit Katz @ Patrk Rećko on Apr 12, 2016 10:22 AM

The little white diamond indicates a refinement event, what this means is that the software splits cells in areas with high velocity gradients in order to get a better resolution. This usually leads to a spike in the RANS discretization error which dissipates quickly in normal circumstances. It seems like in your simulation you are so close to an unstable solution that this spike pushes it over the edge so to speak.

Actions

Patrk Rećko @ Amit Katz on Apr 12, 2016 11:29 AM

Ok, I understand, thank you
So, is there anything I can do to make this solution stable and finish it normally ?

Actions

Amit Katz @ Patrk Rećko on Apr 12, 2016 12:07 PM

Actions

Patrk Rećko @ Amit Katz on Apr 12, 2016 2:41 PM

Nice, I think it worked. Thank you so much! I turned off the automatic refinement, travels, set criterion to stop to all goals satisfied, 8300RPM's on true scale model and enabled cavitation and solver finished normally after 2455 iterations. Now I think I need to rerun calculation on a bigger computional domain and much better mesh, because result plots look strange. Have a look on attached photos.

Actions

Amit Katz @ Patrk Rećko on Apr 13, 2016 8:42 AM

The next step now would be to create a higher resolution mesh (I recommend using manual mesh criteria in your rotating region local mesh) and seeing how those results stack up.

Actions

Patrk Rećko @ Amit Katz on Apr 14, 2016 4:27 AM

I finished the simulation on a better mesh and bigger computional domain, it takes about 5100 iterations and 30 hours, but results are as I expected. I think that problem is solved
Thank you so much for your help and time.
Best regards,
Patryk

Actions

13 Replies

Amit Katz Apr 11, 2016 9:24 AM
Mark Correct
Correct Answer
This error is a catch-all error that means "I give up!" for the software. From how you describe your parameters spiking I'm guessing the solution diverges and the model becomes unsolvable. Possibly the mesh is too coarse. And if you need more than 4500 iterations to solve your model, then it's probably not set up correctly. Any chance you can upload your project?
Like • Show 0 Likes0
Actions
- Patrk Rećko @ Amit Katz on Apr 11, 2016 11:51 AM
  Mark Correct
  Correct Answer
  Zippyshare.com - 4000RPM cavitation ON.rar
  There is the model with 5000 RPM's set up on impeller (name of folder is wrong), I uploaded it with that invaid result, maybe that will help. I noticed that it doesn't depend on iterations, it always getting an error after 2.1-2.2 travels. Weird thing is that I can solve this case up to 3000RPM even on true scale model but if the value is higher - solver terminates. Finally I want to do this simulation with 8300 RPM's.
  1:2 scale model with 4150 RPM and 8 points automatic mesh finish normally after 8900 iterations (internal analysis).
  Like • Show 0 Likes0
  Actions
  - Amit Katz @ Patrk Rećko on Apr 11, 2016 12:29 PM
    Mark Correct
    Correct Answer
    If you save your file without the CFD results then you can upload it directly to the forums. In general to help you we don't really need your results, just the setup.
    Like • Show 0 Likes0
    Actions
    - Patrk Rećko @ Amit Katz on Apr 11, 2016 12:45 PM
      Mark Correct
      Correct Answer
      There is only model with setup
      Attachments
      5000rpm.rar19.4 MB
      Like • Show 0 Likes0
      Actions
      - Amit Katz @ Patrk Rećko on Apr 11, 2016 4:31 PM
        Mark Correct
        Correct Answer
        The file was 130 MB and I was able to load your CFD results, so it looks like you copied everything in. For future reference you can use the "Pack and Go" feature to save an assembly into a zip file and choose whether or not to include certain things like simulation results.
        Like • Show 0 Likes0
        Actions
  - David Paulson @ Patrk Rećko on Apr 11, 2016 6:38 PM
    Mark Correct
    Correct Answer
    Patryk,
    I was not able to open your download file, but from your experience the solver works on a 50% +/- scale model but not on a full scale model at about 5000 RPM. Possible issues:
    1. If the impeller diameter or RPM is not scaled you will not be seeing the same fluid velocities? A 50 mm impeller scaled to 25 mm will produce much lower pressures and velocities at the same RPM.
    2. You did not state the number of cells in your simulation and the comparison between full scale and 1/2 scale. I have had similar issues with simulations with 4-5 million cells, until I increased the DRAM in my computer. More DRAM seemed to fix my problem.
    3. 4500 iterations seems like a lot. Solver is struggling for convergence. Did you, or can you, display the solver solution graph?
    Like • Show 0 Likes0
    Actions
    - Patrk Rećko @ David Paulson on Apr 12, 2016 4:13 AM
      Mark Correct
      Correct Answer
      
      There is less iterations because of reducing the size of computional domain and poor mesh. If there is more iterations the section on graph between 200 and ~450 is longer and almost straight. Solver always terminates a while after this little white dot appears on "iterations axis" and before it appears solver actually achieving all of the goals at 100%.
      Like • Show 0 Likes0
      Actions
      - Amit Katz @ Patrk Rećko on Apr 12, 2016 10:22 AM
        Mark Correct
        Correct Answer
        The little white diamond indicates a refinement event, what this means is that the software splits cells in areas with high velocity gradients in order to get a better resolution. This usually leads to a spike in the RANS discretization error which dissipates quickly in normal circumstances. It seems like in your simulation you are so close to an unstable solution that this spike pushes it over the edge so to speak.
        Like • Show 0 Likes0
        Actions
        Patrk Rećko @ Amit Katz on Apr 12, 2016 11:29 AM
        Mark Correct
        Correct Answer
        Ok, I understand, thank you
        So, is there anything I can do to make this solution stable and finish it normally ?
        Like • Show 0 Likes0
        Actions
        Amit Katz @ Patrk Rećko on Apr 12, 2016 12:07 PM
        Unmark Correct
        Correct Answer
        It seems to me from your goals chart that your critical parameters are more or less converged before the automatic refinement, is there a particular reason you need so many iterations? You could also go to "Calculation Control Options" and manually disable the refinement.
        Like • Show 1 Like1
        Actions
        Patrk Rećko @ Amit Katz on Apr 12, 2016 2:41 PM
        Mark Correct
        Correct Answer
        Nice, I think it worked. Thank you so much! I turned off the automatic refinement, travels, set criterion to stop to all goals satisfied, 8300RPM's on true scale model and enabled cavitation and solver finished normally after 2455 iterations. Now I think I need to rerun calculation on a bigger computional domain and much better mesh, because result plots look strange. Have a look on attached photos.
        
        Like • Show 0 Likes0
        Actions
        Amit Katz @ Patrk Rećko on Apr 13, 2016 8:42 AM
        Mark Correct
        Correct Answer
        The next step now would be to create a higher resolution mesh (I recommend using manual mesh criteria in your rotating region local mesh) and seeing how those results stack up.
        Like • Show 0 Likes0
        Actions
        Patrk Rećko @ Amit Katz on Apr 14, 2016 4:27 AM
        Mark Correct
        Correct Answer
        I finished the simulation on a better mesh and bigger computional domain, it takes about 5100 iterations and 30 hours, but results are as I expected. I think that problem is solved
        Thank you so much for your help and time.
        Best regards,
        Patryk
        Like • Show 0 Likes0
        Actions

Jet ski pump solver terminated

Outcomes

Attachments

Related Content

Recommended Content