Low Supply Air Temperature

[Dymalica]

If a person requires a data room to be 60-65 degrees instead of around 70 degrees, how would this be achieved? The equipment load is constant and there is no wall loads because room is inside building. If the supply air temperature is decreased, this should increase the AC required? If using the formula for BTUH=1.08xCFM(deltaT). The delta T is the same whether it is 75-55 or 65-45. Am I running into a thermodynamic problem with the refrig? Am I using the wrong formula?

 

[MintJulep]

If both sides of a wall are the same temperature, then you have no room load associated with that wall.

If one side of a wall is colder, then you do.

Do you really have no latent heat to deal with at all?

 

[marcoh]

There will be a difference in heat transfer through the wall as the temperature difference will change, but the equipment heat gain should be much larger. Within the limits of the cooling coil ability you should be able to change the temperature with little chane in actual cooling required.

 

[cry22]

Increase your air changes per hour.
You're looking at around 25ACH to get 65F indoor.

 

[DRWeig]

Just my experience and opinion, but when IT folks specify that the data room needs to be 60F, most often they're wrong. They're trying to compensate for hot spots, which is the problem that should be addressed. Today's IT equipment is designed to run in normal space temperatures (70F or so), and can survive excursions up into the 85F range.

Doing 60F space temp can get you into the frosty-coil area pretty easily, ruining the reliability that the cold temperature was supposed to enhance.

Look at info from APC, how to design air flow through data centers / hot aisle, cold aisle concept, spot cooling solutions, etc...

Good luck with it! Let us know how it works out.

Goober Dave

Good on ya,

Goober Dave

 

[mechanicaldude]

cry22, How did you calculate your air changes without know the room volume ??

 

[cry22]

I've done a few hospital OR's (9-10 foot ceilings) with doctors asking for 65F temperatures.
I actually assumed the average office space of 9-foot ceiling.

Model it on Trace or E20-II, just lock in the supply air at 53 to 55F.

 

[imok2]

Good info here:

http://docs.sun.com/source/816-3537-10/2__EnvReq.html

For air changes go here:

http://www.engineeringtoolbox.com/air-change-rate-room-d_867.html

 

[ChrisConley]

It doesn't take more energy or capacity to keep the room cool if it has no external loads. The only loads you need to worry about are the computer equipment.

However, in fact though you've created 'partion' loads as each of the four walls, the ceiling and the floor are now sources of heat from adjacent rooms kept at warmer temperature.

Fortunetly, the temperature differential is quite low. A typical uninsulated partion wall has an R-value around 2. Your temperature differential is 5-10 deg F your area is your walls is a known.

For 1000 sqft of wall this works out to be 10,000 to 20,000 Btu/hr. I'm guessing the extra 1-2 tons of cooling isn't going to make a huge difference compared to the diversity of the servers.

If you are tight, you could insulate the walls.

One other thing to watch is that you will have colder walls on the outside of the server room. Depending on your design conditions you may want to think about condensation.