Air flow pattern from vent

[fidran]

A few years ago, I created a simple FDS model (FDS 6.7.5) that simulates the supply of air from a wall vent (diffuser). This was compared with laboratory experiments, and I found that there was reasonably good agreement between the simulation and the lab tests.
In version 6.10.1, however, I get a flow pattern that differs significantly from the laboratory measurements. It appears that changes introduced in version 6.7.8 are causing this difference.
I have tried adjusting various parameters, but this does not significantly affect the results. I have also tested two different grid resolutions – 3×10 cm and 3×5 cm. Increasing the resolution does not improve the results; rather, the opposite seems to be the case.
Is there anything I can do to influence the flow pattern in the right direction?
Enclosed, lab test and simulations
Air flow pattern.pdf
.

[drjfloyd]

Could you provide your input file for the case?

[fidran]

Air flow (160m3/h) was measured in the inlet duct. Inlet velocity 1.5 m/s (30cm from vent front) was measured right before test start. Air supply temp and room temp was logged every second. The main purpose of these studies was to see if we could calculate the centerline velocities by study the fog movement in the videos. Distance between the black lines on the back wall is 1m. Worked when compensated for camera angle and horizontal fog expansion

[fidran]

[fidran]

I guess your DNS grille is inside a duct - put it on the end of the duct with some space outside, and I guess you will see something else

[drjfloyd]

This looks like the flow one would expect for a negatively bouyant jet; the arc looks parabolic.

[fidran]

I do not disagree that the air jet may have been slightly under-tempered. I am attaching two other tests with the same device and the same airflow. The labels indicate approximately ∆T +5 / -7.
Auranor.pdf

[drjfloyd]

The +5 image shows a jet that hugs the ceiling. The isotherm shows a jet that impinges the floor near the far wall. The -7 image shows a jet that impinges the floor near the midpoint of the room. I ran with the 0.3 m x 0.1 m vent and 160 m3/h at 5 cm resolution. This is still only 6x2 cells for the vent and not very well resolved. Results are below from left to right as +5, 0, -1.25, and -7.

I kept the side walls and ceiling flat; however, the photos show something attached to make a grid. Not account for this is likely having an impact.

Looking at the image of the diffuser there is a round duct whose diameter looks to be ~1/3 the height terminating into a box that I assume is just an empty box with maybe a damper. If that is the case, my guess is the flow profile on the face of the vent is not a top hat profile but biased towards the central portion which does suggest that a reduced area on the wall is a better approximation but still likely not what the actual flow was.

The +5 is to close to the ceiling and the -7 is impacting at ~1/3 the distance rather than ~1/2. The overall trends are in line with the images given the still fairly coarse grid for the vent.

[fidran]

The diffuser has a channel at the rear with a diameter of 160 mm. I am fairly certain that the chamber is empty. In the ceiling there are aluminum profiles with a 60x60 cm grid. The grid on the rear wall is probably 2 mm thick aluminum rails. I believe the ceiling panels are made of compressed mineral wool, surface-treated with white factory paint. They were then painted black afterwards. The floor has a vinyl covering. Heat loss and heat gain from the air jet were compensated for, but I do not remember exactly how this was done in this test room. Possibly an electrically conductive floor covering was used for heating.
I found some temperature measurements indicating that the isothermal experiment has a supply air temperature 0.4–0.5°C below the room temperature. Is the only change in your model that you have added roughness to all surfaces and have a 5x5x5cm mesh?

I have a question related to N_EDDY, L_EDDY, and VEL_RMS. The user manual states that these parameters should be specified on the &VENT line. Is it possible to implement these parameters if you use the following setup:

&OBST XB=xx,yy zz, SURF_IDS='V1','INERT','V1' /
&SURF ID='V1', VEL= -15 /

[drjfloyd]

Thanks for the video.

The synthetic eddy method can only be done on a VENT line. It is not a SURF input. For this method to work, the vent needs to be resolved enough for it to contain eddies.

[fidran]

"I have to leave this for now (tried with a resolution of 3*2.5cm). I am posting the video of the “isothermal” jet. The fog cloud takes about 6–7 seconds to reach the 6m mark. The typical spread angle (horizontal) is between 25 and 30 degrees for such a jet. The jet profile has a typical bell-shaped curve, corresponding to thermal plumes from convective heat sources."

Test.Auranor.TLK.isoterm.mp4