screen with solid particles

[Nadezhda]

Hello, can you help me?

I tried create 2D screen with solid particles as describe "18.4.9 Screens" in User Guide.

If I set N_PARTICLES_PER_CELL=1 as describe in User Guide then air dosen't leak through screen:

If I set N_PARTICLES=800 (total particles number equial in both case) then air leaks through screen:

I attached both input files.

Can you please tell me there my mistake? How should screen work?

[dr_jfloyd]

Add CELL_CENTERED= .TRUE. to the INIT line.  Also set XB(1) and XB(2) to be slightly within the grid cell. By setting it to the cell boundaries you are getting two planes of particles.  To do the "wall of wind" method you need to used SURF with VEL  (see 13.3 in the guide).

[Nadezhda]

Thank you for your answer!

I set CELL_CENTERED and corrected XB but results still the same.

I don't understand why wall wind method will be better but I tried it too.

N_PARTICLES=800, air leaks through screen:

N_PARTICLES_PER_CELL=1, air stiil doesn't leak:

четверг, 6 декабря 2018 г., 22:56:18 UTC+5 пользователь dr_jfloyd написал:

[dr_jfloyd]

N_PARTICLES=800 puts 800 particles randomly in the XB. So you will have random cells with no particles, cells with 1 particle, cells with 2 particles, etc.  Cells with no particles will not put any drag on the wind field.

You have a screen across the entire face of the domain and you have open sides.  The wind is escaping off to the sides. Run the attached case where the screen is in a room where the wind can also go around the building.  In this case you see wind is being forced through the screen as seen in the U-VELOCITY slice below.

[Nadezhda]

Thank you for your answer but I stll have questions.

I slightly changed your file and create two variants: FREE_AREA_FRACTION=0.9 and FREE_AREA_FRACTION=0.2

I think 0.9 should be more "transparent" for wind and 0.2 conversely. But I can't understand results.

 FREE_AREA_FRACTION=0.9 work as wall (as it was in my previos files. But I just made screen bigger):

 FREE_AREA_FRACTION=0.2:

1 second after simulation start

12 seconds after simulation start

It seems like screen creates wind itself.

I can't understand there are my mistakes and why screen works like that.

пятница, 7 декабря 2018 г., 17:50:23 UTC+5 пользователь dr_jfloyd написал:

[Nadezhda]

Have anyone any idea about my problem? Please)

воскресенье, 9 декабря 2018 г., 22:05:10 UTC+5 пользователь Nadezhda написал:

[dr_jfloyd]

I am in the midst of multiple end of the year deliverables and have not had time to look further into this.

[Kevin]

We found a bug in the code related to this. There is a mechanism that limits the time step in cases where a SCREEN has a small opening area. If the time step is not decreased, the force term in the momentum equation becomes so large that it causes spurious behavior. With the current version of FDS, this mechanism is not working properly. As a work around, set DT on the TIME line to a small value, in your case, maybe 0.001. You should then see realistic behavior. Let us know.

[Nadezhda]

Thank you for your answer and sorry it took me so long. 

I set DT=0.001 and now there are no wrong vectors in screen. 

But still there are no wind inside room. I set FREE_AREA_FRACTION=0.99 I think screen sould be almost open but there is no wind inside:

пятница, 14 декабря 2018 г., 16:55:45 UTC+3 пользователь Kevin написал:

[Kevin McGrattan]

It appears that the linear term in the expression for the pressure drop across the screen does not go to zero as the FREE_AREA_FRACTION goes to zero. I have to look into this more closely. 

Jason, do you know why this is happening? 

[dr_jfloyd]

The screen drag model is a correlation for window screens. Window screens do not have a FREE_AREA_FRACTION fraction approaching 1 more like 0.3 to 0.6. Using the correlation in that range would be an extrapolation; I wouldn't expect it to behave properly. 

[Kevin McGrattan]

Jason, thanks. Could you provide a reference and range in the User's Guide.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/fds-smv/e6dc5d7b-5304-49c3-aa1f-870fa20e0c89%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[dr_jfloyd]

Yep, I'll add the reference from the Tech Guide and a note on the range.

[mtd...@gmail.com]

I have previously looked at this correlation for approximating a “porous” false ceiling, say 60% open (thinking it might be used like a distributed porosity “Darcy” type model). But since the FDS manual indicates it is for window screens I have been reluctant to use in anger. Some testing I did showed reasonable results only with very low blockage/pressure drop. Can you comment on the circumstances in which this model might be used? Any other options for representing “porous” false ceilings in FDS, besides trying to appproximate with obstructions?

[dr_jfloyd]

There is the general porous media model where you can set the K and Y parameters for the drag. Follow the screen drag verification case setup and make a similar model at the grid size you wish to use and tune the Y and K to get the correct pressure drop. For a perforated metal ceiling panel the linear portion of the drag term isn't very important so you could make K "large" and just tune Y.

You could do the same with the screen model. At the time we created the inputs for the FM/FPRF datacenter validation cases we didn't have the general porous media model. If you look at the current FM/FRPF datacenter validation cases, we do use the screen drag model to represent a perforated tile in a ceiling. But you can't just take the panel thickness and free area. The approach taken was to start with the screen drag verification cases and keep the MESH IJK but change the XB to get the same resolution as the validation case. Then the free area fraction was set to the perforated ceiling and the thickness of the screen was adjusted until the correct pressure drop was seen. We should probably change those inputs to use the general model now that we have it.