firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
673 stars 627 forks source link

implementation of U0/V0/W0 #13696

Closed ericvmueller closed 3 weeks ago

ericvmueller commented 3 weeks ago

Reading the guide I believe the intent of U0,V0,W0 on WIND is to initialize the flow field to reach a steady state faster when using a VENT to drive the flow:

image

However, it seem like if this is used in combination with a vent there is an additive effect on the VENT velocity and the flow still takes time to establish. For example, run the lee_waves.fds case. The input is

&WIND U0=5.0,LAPSE_RATE=0.005 /
&SURF ID='WIND',VEL=-5.0,PROFILE='ATMOSPHERIC',PLE=0. /

The inlet flow starts at 5 m/s but then accelerates to 10 m/s

image image

mcgratta commented 3 weeks ago

I think that this advice is outdated. I suggest you remove it from the User's Guide.

ericvmueller commented 3 weeks ago

Is there an accepted way to start with a flow field that is matched to a vent velocity - which I think the intent is with that lee_waves.fds case?

rmcdermo commented 3 weeks ago

Could you use RAMP_UZ? Have a look at the user guide.

ericvmueller commented 3 weeks ago

I still get an additive effect. In my case I have

&MISC RAMP_UZ='u'/
&RAMP ID='u', X= 0, F=2 /
&RAMP ID='u', X= 1, F=2 /
&SURF ID='INLET', VEL=-2/

I start out with 2 m/s like I want but then things accelerate to 4 m/s

image image

mcgratta commented 3 weeks ago

Have you looked at INITIAL_SPEED?

ericvmueller commented 3 weeks ago

I get the same as above. Iooks like INITIAL_SPEED is just another way to define the U0 and V0 global variables

mcgratta commented 3 weeks ago

I changed the code so that now, INITIAL_SPEED and INITIAL_DIRECTION (new parameter) do nothing except set U and V at T_BEGIN.

ericvmueller commented 3 weeks ago

Try with this input - I still see the same issue. Unless I'm just not understanding the expected behavior.

initial_speed.txt

mcgratta commented 3 weeks ago

Yuk. I think FDS is ramping up the specified normal velocity assuming INITIAL_SPEED is the ambient. I'll fix.

mcgratta commented 3 weeks ago

I remember now the problem with initial velocity. In the past, we would have to subtract off the initial velocity components from calculations of $\partial{u}/\partial{t}$ etc. This became a burden, as this was done throughout the code. It seems that INITIAL_SPEED is no longer working and not used much. To make it work would involve going back to all these subtractions. I don't think it is worth it. I removed it completely.