| |||||||||||||||||||||
|
Beware of British Thermal Units in the United States Members of the ISP-Tech list share elaborate solutions to the simple problem of calculating how to keep your valuable racked equipment cool. But beware, BTUs can become big headaches.
On the ISP-Tech list in August, AM asked,
A number of respondents shared their experiences with the challenges of air conditioning: [SD offered] "We have generally found that if your computer room is usually unattended (i.e. no one permanently sits in there), humidity won't be as much of an issue as temperature: the biggest producer of humidity within an office is people. If your computer room doesn't have windows, humidity will generally be low. In terms of units, the cream of the crop is Liebert (www.liebert.com); they're specifically designed for 24x7x365 operation in computer rooms-but they're very pricey. The next best thing is a ductless wall-hung unit: refrigerant is pumped right into the unit from the compressor, and the evaporator/blower 'hangs' on an interior wall. They're usually quite reliable, easy to take with you if you leave, and are 35-50% cheaper than a Liebert-but they're not really in the same league." [RT added] "We currently have two Mitsubishi wall units: they've served us well. The datacenter has held a nice 70 degrees since inception. The most important considerations are:
In general, go with cheap, easily replaceable, and redundant. Then again, that's my view on most things ISP-related: even the best-managed, best-maintained, lowest-failure-rate equipment WILL fail..." Others suggested some methods of calculating BTU requirements: [KE advised] "If you can calculate your amp draw, you should know how much 'heating' you have going on. I think a unit between 8000 and 10,000 BTU ought to do it. If humidity remains high, then you can add a dehumidifier. If you go too high in BTU, the room will be cold but may still be humid." [GG countered] "The industry standard would be a two-ton (24,000 BTU) AC unit; the standard for a computer room is one ton per 55 square feet. The other thing that is standard in a small room is to run the fan in the unit continuously, as recovery time my be too high otherwise." DV complained that the methods of calculation are way too complicated: "Why don't they rate these things in Watts? I know how much power I'm feeding into the room in Watts; I therefore know how much heat I need to take out of the room in Watts. There's probably a conversion formula somewhere: tons is totally archaic, kinda like 'furlongs per fortnight'…" Still others responded with some advice: [GG offered] "There is a formula; the problem is, it only works for resistance heating systems. There is no direct conversion from Watts to BTU, as different items can use the same number of Watts to generate different amounts of heat. If you look at the detailed specs on each piece of equipment you have, one of the items specified is heat dissipation in BTU. Make sure you have cooling equal to the total heat dissipation of all equipment in the room (including lights, etc.), plus an additional 4,000 BTU per 100 square feet at a minimum. In a fully equipped room, this normally figures out to be 12,000 BTU for every 50 to 55 square feet. Most engineers shortcut the process and just use the formula of one ton, or 12,000 BTU, per 55 square feet." [BS added] "Factors such as the stabilizing effects of insulation, a cement floor, etc., can also be entered into the equation. There's a real science to this stuff. A competent specialist can remove the guesswork from this particular equation: the money you'll save by calculating the specific requirements may make it possible to install a redundant system or other bells and whistles. Finding someone that knows AC on this level may be difficult, but it's certain to be worth the effort." [MF noted] "You can check out conversion calculators and formulas at the following websites: Energy and Power: units and definitions"
End
|
|
|||||||||||||||||||
|
|
|||||||||||||||||||||
#