inlets/outlets Not created automatically/Not enough memory to run radiation

[Anonymous]

Autodesk CFD Forum;

I have a rather complicated largescale model 1 of 24 section with one curvilinear and 3 flats.  Put a 100Pa pressure BC on the "inlet" and a 0 velocity reverse normal on the "outlet".  Using variable air material on atmosphere model and set glass as the material on the greenhouse section.  Using IV2016 -)> CFD 2016 to analyze.  Shouldn't it have created an inside volume and automatically set the inlets and outlets for the velocity fields IAW the BC's ?

I've tried the window material to no avail on the larger curvilinear surface. 

Dell Precision T7500 8 core xeons 96GB RAM  ... When running the radiation experiment I run out of memory according to CFD at 99% iteration point yet my monitor shows only ~18GB used.

I need to see these flow patterns and not once of dozens of tries has CFD reported back inlets/outlets in the messages and I don't think it is creating the special internal volume.  Do I need to run the IV shell command on the greenhouse section?

Any help in this matter will benifit us greatly.

Thanks in advance,

R. Harker

Messages

==============================================

Mesher process started...

Loading model...

Loading mesh sizes and constraints...

Meshing...

Surface meshing... generating surface mesh

Surface meshing... smoothing surface mesh

Volume meshing... creating volume mesh

Volume meshing... optimizing volume mesh

Volume meshing... smoothing volume mesh

Mesh contains 664830 nodes and 3466762 elements.

Performing mesh quality check...

Saving mesh...

Meshing completed.

Updating model database...

Model database updated.

Updating mesh database...

Mesh database updated.

Building solver model file...

Solver file written.

Initializing ...

Control file processing complete

Check Out motion License: Successful

Authorization verified

Processing input ...

Mesh input complete

Processing results input ...

Results input complete

Processing BC input ...

Computing restart BC data ...

Calculating BC data structures ...

Calculate wall distance for 664269 fluid nodes

Wall distance calculation done in (3 + 8) seconds. Max distance = 3.09835e+06

BC data structures complete and saved

Analysis Initialization Complete

Input processing complete

** FINITE ELEMENT SUMMARY FOLLOWS...

664830 Total Nodes, 664269 Fluid Nodes , 561 Solid Nodes

3466762 Total Elements, 3462177 Fluid Elements , 4585 Solid Elements

0 Inlets 0 Outlets 0 Unknowns

*********** ANALYSIS STARTED ***********

Turbulent Compressible Flow is ON

Turbulence model: Standard k-epsilon

Intelligent Wall Formulation: ON

Flat lines detected in Convergence Monitor - Auto Stop.

Begin Post-Processing Calculations ...

Post-Processing Calculations Done

Begin Post-Processing File Output ...

Post-Processing File Output Done

Analysis completed successfully

[TravisNave]

Technically there is no memory limit to CFD.  But it is possible that your computer is not configured properly to use it all.  Are you sure you can see all 96GB of memory in the computer?  I know that I have a server with that much memory that was limited because of a virtual machine limitation, so verify that Windows can see it all.  Also, what is your virtual memory swap file set to?  If it is too low to swap out memory, then that could also be a problem.  You'll probably want to set it to at least 1.5x-2x the physical RAM and NEVER let Windows manage it.  I would set the min and max to 196608 just to be safe.  You'll obviously want to have lots of HDD space as well for this VM and working temp files.

Good Luck!

[Anonymous]

Travis -- You were correct that windows was managing the cache ... I think that may solve the radiation memory fail but I haven't tried it yet.

I tried creating a thickened object using the shell command in IV and CFD did automatically create the internal volume.  Running it now to see if it converges for a simple flow sim.  However I'm still seeing "0 inlets 0 outlets 0 unknowns" in the CFD message window so I don't think it will work.

Another quick question:

I'm not able to skip through the interations after a successful simulation (didn't diverge).  IDK how or I have some parameter set wrong that doesn't save each sim interation.

Thoughts?

Thanks,

[Anonymous]

And it didn't work ... flatline convergence interation #3

[Anonymous]

@TravisNave and anyone else -- My greenhouse section is begining to behave somewhat but the velocities are WAY off.  The outlet flows should be in the order of meters/sec but the analysis reports 20M - 40M m/sec and I'm only seeing an output at the exit surface and no internal flows.  Is it because the external volume I have should be much much larger and the variable air in this volume is heating up the whole external space?

I was able to use the surface wrapper to create an airtight surface of the section whereby I could select my BC's for inlet and outlets by the traditional method of seting a pressure on one and a velocity on the other.

Still am not seeing the solver recognizing the inlets/outlets but in this run at least the outlet seems to be working albeit like a rocket:

0 inlets 0 outlets 0 unknowns

So it's begining to look like a simulation but it has problems.

Thanks again;)

Robert

[Anonymous]

PS: Should I be using a much much larger dome shaped external volume?

[TravisNave]

Could be. Wish I knew more about it to help you out. Do you have
Subscription support? Might create a support incident if you do.

[Anonymous]

@TravisNave Not on subscription.  Using Autodesk's wonderful educational offerings ... which we REALLY REALLY appreciate.

It's a model section of a solar chimney (tower) (google) ... quite a facinating and ascendent technology that I have be working with for more than 20 years.

Personal growth is what it's all about and we are thankful.

Robert

[Jon.Wilde]

Hello,

Hope you don't mind me adding a few points here.

Firstly, test the model without radiation if you can.

Looking over your screencast I see a few things:

1. What does the velocity=0 represent? I think I can say that I have never applied this 🙂
2. Are all of your boundary conditions on external model boundaries? They do need to be - near the end of the video it looks like they are internal - this will not work. If you need to model openings, just model the solid part and leave it open to the air
3. Your flow does not look smooth, which typically means that your mesh is insufficient
4. Your domain is 10km tall, is that right?
5. You need some sort of ambient temperature somewhere, or nothing can happen as CFD has no baseline - you will likely see just three iterations. Did you run through any of our tutorial models or see the guide here?

I hope that is useful.

Try running with a heat load on the volume of the chimney or something initially, so that you can have flow moving and have radiation off. Often is it just a complication that causes issues until the basics are all set and working.

Thanks,

Jon

[Anonymous]

@Jon.Wilde

Hello,

Hope you don't mind me adding a few points here.

I'm a CFD newb ... please point away!

Firstly, test the model without radiation if you can.

OK -- I have tried a few runs without radiation and have gotten flows before.

• Looking over your screencast I see a few things: What does the velocity=0 represent? I think I can say that I have never applied this :

That would be the inlet BC.  The velocity BC was set to the outlet.  I read in here somewhere that one needs to set a pressure and a velocity to get a flow.

• Are all of your boundary conditions on external model boundaries? They do need to be - near the end of the video it looks like they are internal - this will not work. If you need to model openings, just model the solid part and leave it open to the air

I'll try that Jon.

• Your flow does not look smooth, which typically means that your mesh is insufficient

The flows looked smooth but they were rocket engine velocities. 

• Your domain is 10km tall, is that right?

Yes ... solar chimneys are very big if they are built right.

• You need some sort of ambient temperature somewhere, or nothing can happen as CFD has no baseline - you will likely see just three iterations. Did you run through any of our tutorial models or see the guide here?

I have done some of the tutes. So I should set an initial condition of temp =25 o C for instance of my bounding volume and my internal air volume correct? Does variable air act as an atmosphere at 10Km?  Pressure and temp gradients? 

I hope that is useful.

Very much so.

Try running with a heat load on the volume of the chimney or something initially, so that you can have flow moving and have radiation off. Often is it just a complication that causes issues until the basics are all set and working.

Rodger that.

Thanks,

Jon

====================================

Very informative ... thanks;)

Robert

[Jon.Wilde]

Feel free to share your next CFZ when you are ready.

Remember to just leave parts of the chimney open if they are open in reality, no need to prescribe anything like pressures or velocities if natural convection takes care of it anyway 🙂

I think I would just set an ambient temp on the top surface for now. Also think about the ground, if you have solar loading, you might need to model a thickness of soil and add a film coefficient beneath it (as the ground will absorb some heat too!).

[Anonymous]

What does it mean when the CFD solver generates RED text? With this iteration of the model I keep getting solver errors. Should I abandon the 500mm shell and just use the original solid shape?

Thanks Jon;)

UF Thanks you.

[Jon.Wilde]

Hi Rob,

Is everything Red or just some items? What are they?

I would use a solid, yes - usually far easier to work with and more accurate here as you have solar loading and really need a good heat transfer through the part also 🙂

Thanks,

Jon

[Anonymous]

Dear Autodesk CFD (jon);

Is everything Red or just some items? What are they?

All the mesage window messages were colored red ... untile the solver stopped

I would use a solid, yes - usually far easier to work with and more accurate here as you have solar loading and really need a good heat transfer through the part also 🙂

I've got one more shell I'd like to run with no radiation ... simply 75W/meter apllied to the earth shape simulating a good days radiation to the ground although we get upwards of 100W/Meter. I also put 70W/meter inside the greenhouse floor to simulate the attenuation from the glass losses.  I've had issues with both methods (shell vs. solid).  And the shell method just failed again.  So back to the solid form.  I have simpler models to run as well.

This personal CFD study is tough on an old engineer.   But I am progressing.  I'll screen cast the sucesses when I start seeing them.

Thanks Jon,

Robert Harker

[Anonymous]

Jon;

Running using solids imported to CFD but I made the mesh exceedingly tight since I want to see small features. My workstation is almost maxed out at 96GB but the 8 cores are NOT and I'm already saving the mesh.... you will see in my next screencast.

Wish me luck ...

Rob

[Anonymous]

I think I just overtaxed my system with the small mesh size.

I'll do better.

////(--)\\\\

RobHarker

[Anonymous]

I'm still not getting the internal volume as much as I tried several methods but this sceen cast shows more action than any other run with reasonable velocities.  (I'll show the screen cast later when your servers allow me)

The forced covection is working but is there a natural convection if I leave it unchecked?

Should I delete the faces in Inventor of the inlet and outlet to create a natural inlet/outlet?

I think I need a much much bigger volume to simulate the atmosphere.

Also -- still seeing "0 inlets 0 outlets and 0 unknowns" in the message window.

I've never seen those occur in the dozens of models I've run in the last weeks.  That is my problem isn't it?

If you have the time please see my screencast.

Thanks in advance.

RobHarker

[Anonymous]

PS: I tried to set the save intervals so I could look back at the iterations and it allowed me to do it in the solver yet I couldn't step back through the results. I know I still have a "solid" weirdly however there are flows from the assigned inlets/outlets at times...???

[Jon.Wilde]

I am on holiday but a super quick reply as we were talking before 🙂

Do remove inlet and outlet volumes from the model, just have the real world solid and an air volume.

Do not assign any internal boundary conditions.

Turn off Auto Forced Convection (this runs flow first and thermal second). It will not work if flow and thermal are dependent on each other, like here with natural convection.

Once I am back, I can take a look at the support file (CFZ) and give you more detailed steps 🙂

Hope that helps!

Jon

[Anonymous]

Man working on holiday ... enjoy life;)