HVAC fan definition

[iceandfire12345]

Hello,

I'm using PyroSim (this is still an FDS question) to model and simulate a simple fire burning in a tunnel whose smoke must be extracted by two fans placed a few meters beyond the fire. The smoke is expelled through a chimney with two seperate flues to the atmosphere. There is also a Monin-Obukhov wind blowing in the -x direction.

The tunnel has its own mesh with two open vents at each end; the atmosphere also has its own mesh which has open vents everywhere except for the z_min plane which is inert. Between the tunnel and atmosphere are three meshes, the middle one being the fan (see below).

I modeled the fans as HVAC fans using two HVAC vents (one facing the tunnel, the other the chimeney) and an HVAC duct connecting the respective HVAC nodes placed on the vents.

It wasn't clear from the FDS User Guide, but must there be an obstruction between the two HVAC vents? When I do have this obstruction, the W velocity is positive in the chimney flues, as expected (see below).

However, the obstruction blocks tracer particles from passing through, so I also tried the case without an obstruction and just the mesh (the HVAC vents are at the z_max and z_min boundaries of the mesh). This simulation produced very weird results, where the W velocity in the left flue was positive, while in the right flue it was negative.

I also tested this situation with the left fan turned off and got the same results:

1. What is the correct way to define an HVAC fan? Must there be an obstruction between the nodes? Is there any way to allow tracers to pass through the fans?

2. What do you think caused the anomaly (negative W velocity) described above? Was it just an incorrect definition of the HVAC fans?

Thank you very much!

[drjfloyd]

HVAC VENTs must be attached to a solid surface, but they do not have to be connected by the same OBST.
An exception is if you want particles to flow through the duct. Currently that is only possible when two VENTs are connected by a single duct and the OBST is one grid cell thick. See the User's Guide for deails.

I don't know what your exact inputs are so I cannot comment on why you are seeing what you are seeing. Attach a simple fds input file (not a psm file) demonstrating the issue. You can change the extension from fds to txt to attach to the discussion.

[mcgratta]

WIND is inappropriate in this case. The M-O profiles are applied to the entire domain. If you want some sort of blowing, do it using a VENT with a VEL.

[iceandfire12345]

Do you think that it is the MO wind that's causing the weird graphs I showed above?

I removed the MO wind and indeed the instability in the Fan_test_fire model disappeared. However, the mass flow device worked fine even with the MO wind.

[mcgratta]

The M-O profiles are intended for a large, open domain, not yours. Any good or bad results you get using it in your context are purely fortuitous.

[iceandfire12345]

What is a rule of thumb for how "large" the domain must be? The big model is 100 m x 100 m x 50 m, do you consider that too small?

[mcgratta]

It's not the size, per se. The M-O profiles were designed for a rectangular domain that is open on top and sides and solid on the ground. It is not meant for a case where you want a wind outside of some building. The whole domain must be the outside.