SimScale CAE Forum

Simulation Error


#1

Why am I getting such ridiculous result after I rerun a same simulation?
Project Link: https://www.simscale.com/workbench/?pid=5678998373039009181&rru=78f7cd43-5e69-4d3c-aad8-6734b444a043&ci=377f6dad-7742-4436-b229-7f2450a1c4d7&ct=PLOT&mt=SIMULATION_RESULT


#2

Initially the force plot gives a maximum value of ± 10N only


#3

@PowerUsers_CFD


#4

Hi @100067610!

I (still) believe that the distance of your inlet to the rotating component is way too small and also that the domain in general can be extended to make sure your forces are calculated in a correct manner. That solved quite a lot of issues like yours in the past!

Best,

Jousef


#5

A few things to try:

  1. In CAD file you need to increase the MRF cylinder radius so that it fully encloses your blades, but just a little bigger than the blades :wink: .
  2. While you are in the CAD file, simplify your geometry a little by removing the ‘axle holes’ that the blade spins on. They add unnecessary ‘confusion’
  3. I agree with @jousefm, but if you insist on using that size BMB, then try changing all BMB faces to Custom>Inlet-Outlets with ZeroGradient pressure and set the air velocity of the simulation run inside that Custom BC (I would do both, increase BMB size and use the described BC). Increasing BMB size only adds a few cells as they are generally Level 0 cells …

#6

ok sir, will give it a try now. Thank you!


#7

ok sir, I am limiting the BMB size because I am afraid that it might cause machine to run out of memory. After your explanation, I think I dont have to worry about that anymore! Thanks!


#8

You have 9M cells already, making BMB bigger by 3 or more times will only add a very small percentage of that.

Most of your cells are near the geometry.

You can try a little coarser mesh if you wish, but there are small details that need refining…

Also 3 instead of 5 layers could help reduce mesh size without much results accuracy loss…


#9

Yeah, I have a lot of refinements. Thats why I am getting 9M nodes. Haha


#10

By the way sir, had anyone having the same problem as me before?


#11

You showed a simple set of results from poorly converged data.

Your results were converging, but at that rate, you would use use all of even my core hours to become 1% stable over 500 iterations :wink:


#12

Hello @100067610,

Nice design of a compressor using Savonius wind turbine! Mesh is now quite big despite ‘Very Coarse’ selected for automatic meshing, so it is not easy to put hand on it here, on world’s edge! :lizard:

In my opinion BMB is too big (too long), but I’m also used to simulate ‘free air’ devices with wall defined as ‘Custom’ and all other elements for that wall set to ‘Zero Gradient’.

When you run a simulation it is perhaps good to have possibility to visualize intermediate steps of developed stream, turbulence, etc. This can be done with Simulation Control > Write Interval. If you let you simulation run for 500 seconds, set Write Interval to 50 seconds. That way you will have 10 ‘windows’ into your simulation.

Now, back to simulation, it is definitely a complete disaster and having such a high values for forces is the first warning.

Using your cutting plane I can see X speed from -50 to +20 km/s, which is quite impressive. You see your Savonius turbine compressing air against the flow direction.

Next time, when you see it yourself, you can try to mend it by revising your assumptions.

Simscale, as CADA (it was @DaleKramer idea): Computer Added Design Analyses, is a very good tool to play with your ideas. Now you need patience and (good) results will come. :footprints::man_student:

Cheers,

Retsam


#14

Sir, I am actually doing some comparison for my designs. For this project link , the result is totally different from this huge value one. The changes in geometry is only a little. I dont know what went wrong :frowning:

https://www.simscale.com/workbench/?pid=3012495203542830227&mi=spec%3Ab42d91df-a274-4e00-b397-0816995a2886%2Cservice%3ASIMULATION%2Cstrategy%3A1


#15

The wind speed input is only 5m/s , I dont know why it became so big when the simulation finished


#16

Welcome to CADA, sometimes you never know :wink:

The smallest thing can cause divergence…

EDIT: I have even had a set of results that become stable and when I re-run with exactly the same sim parameters, the results diverge, go figure…


#17

Until you get stable results do not be concerned that they are big, when they get stable they will be small…


#18

Anyway , I am still going to try out what @jousefm suggest to me. Hopefully it shows a different result. Feeling so stressful now :frowning_face: haha


#19

Feel free to try it and let us know if you would need more cores but test it first, the PowerUsers have access to 32 cores :slight_smile:

Cheers and keep us up-to-date!

Jousef


#20

@jousefm suggestion will generally be taken care of by the Custom BC that @Retsam and I suggested (and a bigger BMB that I would add too since I still don’t completely trust even that Custom BC as much as @Retsam :wink: )…


#21

Sure, he can try that and see what effect(s) both approaches have :slight_smile: Open for suggestions! Good job and nice collab between you and Andrzej as always :wink: #dreamteam

Cheers!

Jousef