search

firemodels / fds

Fire Dynamics Simulator
https://pages.nist.gov/fds-smv/
Other
667 stars 625 forks source link

Data Center Modelling #2228

Closed gforney closed 9 years ago

gforney commented 9 years ago 
Please complete the following:

Application Version:6.1.0
SVN Revision Number: 6.1.11
Compile Date: Oct-08-2014

Hello Developers,

I am looking for some enhancements to solve my model issues. The FDS file in the attachment
is the simplified version of my data center project. The following are the brief description:

1) The server IT equipment are modeled as a heater (kW/m^2) to generate about 1200
kW

2) The air conditioning unit is model as a cooler (-kW/m^2) to remove the heat which
generated from the equipments

3) There are two main types of fans, one is the big supply HVAC fan and the other one
is server fan, but I lumped all server fans together in this simple case. The flow
rate i used is based the steady state energy balance equation  q = density x Vol_flow_rate
x C_p x (T_return - T_supply)

4) The system is suppose to operate in a closed loop (Isolated from the outdoor condition).
The temperature, T_supply, of air to equipment is 20 C
and the return air temperature is 35C 

However, i could not able to simulate this situation. I found the result for the pressure
field is keep going down and can't stable itself and lead to numerical instability.
However, when it is open to the ambient, the pressure become stable, i am wondering
what cause this in the closed system.

The overall average temperature range is also not right range (20-35 C, delta T=15）

Lets me know if you need more information,

thanks for your help,

Shawn, Si Tu
5148482424 ext 7236

Original issue reported on code.google.com by Shawn.situ on 2014-10-09 04:26:09

gforney commented 9 years ago 
You defined INERT walls.  These are walls that are ALWAYS at ambient temperature.  You
add 1200 kW with the heater, you remove 1200 kW with the cooler, and you will loose
some heat anywhere air warmed by the heater comes in contact with your INERT walls.
 Heat losses > heat inputs.  What happens in a sealed domain if you do this?

Original issue reported on code.google.com by drjfloyd on 2014-10-09 11:55:08

gforney commented 9 years ago 
(No text was entered with this change)

Original issue reported on code.google.com by mcgratta on 2014-10-09 13:22:25

gforney commented 9 years ago 
Hi, drjfloyd

Original i thought if the exterior wall is set to inert, it will help to balance the
excess heat or over cool situation.

I have tried the setup with all exterior walls and internal walls to be adiabatic.

The pressure measurements for this case is going up and temperature too.
I also found that both mass flow rate of fans are affected because the density could
not stable(the temperature keep increasing)

Do you have more suggest solution, it seems tiny difference in heating and cooling
will cause error piling up. Some data is plot and shown in attachment

thank you for your help,
Shawn

Original issue reported on code.google.com by Shawn.situ on 2014-10-09 18:34:29

gforney commented 9 years ago 
CFD models are subject to numerical error. That error is strongly dependent on the grid
size.  You have a fairly large grid.  It is not surprising you are seeing some error.
As you have observed, a small error over time can add up.

Constant volume flow only equals constant mass flow if density does not change.

There is no such thing as a perfectly sealed structure. In the real world, any imbalance
between heating and cooling would not pressurize that structure to multiple atmospheres.
At some pressure change, probably measured in 10s of pascals at most, leakage flows
would act to limit further pressure change.

Original issue reported on code.google.com by drjfloyd on 2014-10-09 18:52:51

gforney commented 9 years ago 
Hi drjfloyd,

Follow with your suggestion, I have created four leakage hole on the exterior wall
(front and back), the pressure curve now has stopped building all the way up or down
as before, but cycle in short range.

attachment has my current best fds set up. 

So My next thing trying to solve is the temperature distribution. the current result
shows that the temperature is about 34C going into heater and come out at 52C in average
(excel in attachment). 

If we want to achieve average 20C supply air in and average 35C out of equipment (heater)
and then cool back to 20C in a cycle. Do you have any method to get into the right
range. 

I used q = m_dot x Cp x (T_2 - T_1) equation, it works well in the open system where
the air is coming directly from outdoor at 20 C and leave the heater around average
35 C and exhaust out of building. However its not very well in this closed loop operating
mode. 

regards,

Shawn

Original issue reported on code.google.com by Shawn.situ on 2014-10-09 23:35:28