[Bollonga]

I've chosen standar values of density 1.225 kg/m3 and viscosity 1.7894e-5 kg/ms given by fluent.
I've chosen the timestep following the rule of CFL=1; delta_x/U=1. I even chose smaller ones.
I can try hexa-mesh, but I don't know why in the references they get good results with worse meshes than mine.
I prefer not to use a very fine mesh cause later I'll have to test the real scale case, and that's going to need even finner mesh.
Also in the end I'll have to do the 3D case.

[Far]

If one is getting the good results with bad mesh then he must the lucky one and atleast I cannot believe these results. In CFD one must first ensure the best practices instead of getting close to the experimental data.

[Bollonga]

You are right, but I'm worried about computational time.

[Bollonga]

Originally Posted by Far

Do you wanna try the hexa mesh? I have made one. In this mesh, Y+ is 1 (good for transition model as well) but requires more time steps. You may need the transition model for better prediction of Cd.

Which model would you recommend to use in your mesh? k-w SST? Which time step would you recommend?

Would you pass me the mesh you just made, please?

Is there a tool to estimate computational time required?

Thanks for your help

[Far]

Originally Posted by Bollonga

Which model would you recommend to use in your mesh? k-w SST? Which time step would you recommend?

Would you pass me the mesh you just made, please?

Is there a tool to estimate computational time required?

Thanks for your help

mesh,case, dat and ICEM files are attached here:
https://dl.dropbox.com/u/68746918/PS_2D_30Hx20H.zip

1. I would recommend SST-KW model

2. You should study the time step effects or do some literature survey and see how people have determined the time step.

3. Mesh is attached

4. You can estimate from the time step size.

[Far]

Velocity contours (results seem to be good)...

[cfd seeker]

Originally Posted by Far

If one is getting the good results with bad mesh then he must the lucky one and atleast I cannot believe these results. In CFD one must first ensure the best practices instead of getting close to the experimental data.

Spot on. Thumbs up. You are 100 % right. We should stick to the basics first and apply proper physics to capture the desired phenomenon

[cfd seeker]

Originally Posted by Bollonga

By scaling I mean to reduce the scale of the mesh.
My real plate is 9.13m but I wanted to compare with literature experiments for a 0.15m plate. So I reduced mine to have the same length.
The real scale problem gave TVR limitation but the same mesh with reduced scale gave no problem.

So here you are applying "Similarity Principle", have you also changed the Re. No and Mach No. to get the exact similarity b/w the two cases?

[Far]

Generally we do not apply the Mach number similarity when we are not interested in shock-wave interaction. So we must ensure the Reynolds number similarity.

One more thing, in making such analysis if one is interested in flow parameters change due to periodicity in flow then we one must be clever enough to set the density, dynamics viscosity, velocity and characteristics length in such a way to get the desired results with less man and machine hours. For this we should consult the work done by others in literature (Ansys customer portal is good place to find such data). In this way we can save our time and can use itin extending the previous work.

To just give you example: If you want to simulate the flow around cylinder at the Re = 200. How would you set the parameters: density, viscosity, velocity and dia of cylinder? will you choose the default density = 1.225, default viscosity =1.78e-5 and then decide the velocity and diameter? Any idea even selecting the velocity and diameter? What will be the effect of selecting very high or very low velocity?

PS: If one is interested I can share the Cylinder simulations at Re = 200 (Fluent and ICEM files)

[Bollonga]

@far I don't know how to use ICEM very well, is the mesh already generated in the files you sent me or do I have to specify number of division along the blocks sides?

I couldn't open fluent.cas either.

Did you get any Cd or Cl results?

Thanks a lot

[Far]

Yes ICEM files are ready to be used. You are not able to open the case file becasue these are V 14 files and you are using older version. I am also attaching the .msh file which you should be able to open in any version of Fluent.

https://dl.dropbox.com/u/68746918/fluent.msh

[msaeedsadeghi]

Dear friends,
I have not used ICEM till now. Is it really better than Gambit in meshing?

Regards,
Saeed Sadeghi

[Far]

yeah in most aspects it is better. But some geometry functions in gambit are more powerful.

[cfd seeker]

Originally Posted by msaeedsadeghi

Dear friends,
I have not used ICEM till now. Is it really better than Gambit in meshing?

Regards,
Saeed Sadeghi

ICEM blocking topology for hexa meshing is awesome. The tools in the blocking has made hexa meshing relatively easy as compared to other meshers

[cfd seeker]

To just give you example: If you want to simulate the flow around cylinder at the Re = 200. How would you set the parameters: density, viscosity, velocity and dia of cylinder? will you choose the default density = 1.225, default viscosity =1.78e-5 and then decide the velocity and diameter? Any idea even selecting the velocity and diameter? What will be the effect of selecting very high or very low velocity?

Far you have highlighted a very important point. Can you tell me how you approach the problem in this case if just the Re. No is given in the reference data? I normally use default values for density and viscosity, body length is always known to me. Based on these parameters I calculate velocity and then the mach no and then run the run the case at this calculated mach no? Is there any other method? how you do in such a case?

[RodriguezFatz]

cfd seeker - I hope I didn't get you wrong: If you know the Re number, what is the problem? It absolutely doesn't matter how you "replicate" the Re number, just the number has to be the same. Results will also be the same (in normalized values).

[Far]

I would use density = 1.225 Kg/m3, V = 1 m/s , dia = 1 m and then viscosity = (1.225*1*1)/200. In this case stroul number is 0.22 approx and frequency will also be 0.22. Now I can decide the time step size.

[Bollonga]

Hello everybody,

Sorry, I've been busy for some days. Returning to my problem I've tried to use the mesh Far made for me, but when I export it into Fluent v6 format it shows an error message (see picture). It's related to boundary conditions but I've specified all of them as usual: velocity inlet, symetryx2 and pressure outlet. Is anything left?

By the way, Far, I'm interested in your cylinder Re=200 simulations if you still want to share!

Thanks

[Far]

I am always ready to share the knowledge. https://dl.dropbox.com/u/68746918/ICEM_CFDOnline.rar

By the way why file is giving error? are you using ICEM V14? You can directly read the .msh file in Fluent.


PS : Drop box will take some time to upload the files. May be 10-20 hrs, meanwhile you will get error 404 or 403

[Bollonga]

Originally Posted by Far

I am always ready to share the knowledge. https://dl.dropbox.com/u/68746918/ICEM_CFDOnline.rar


PS : Drop box will time to upload the files. May be 10-20 hrs, meanwhile you will get error 404 or 403

Thanks a lot. How big are the files?

I've managed to export your mesh in 2D, but it gives error when importing it in parallel mode. An error message says:

Build Grid: Aborted due to critical error.

There are also several messages in the command window saying:

4:WARNING: cell id 28340 of thread 24 has NULL face pointer 3.
Primitive Error at node 4: Build Grid: Aborted due to critical error.

Which can be the reason?