Heat dissapated from servers 3

[darcope]

i would like to know the heat dissapated from servers, plotters, CPU's, ect. normally found in Computer rooms.
I found a blurb in a HP website that says "the total wattage input is eventually dissapated in heat", but when i call the CRAC unit vendors such as Liebert, they say the total wattage is not used as heat. which is correct?

 

[Twoballcane]

For design purposes you use total wattage as heat. You can not dissipate more heat then what is going in. However, in reality the whole circuit is not on at the same time like a light bulb. Some parts of the circuit will be on while some other parts will be off and then vice versa, but it is futile to figure out the heat dissipation due to this flip flop.

Tobalcane
"If you avoid failure, you also avoid success."

 

[RossABQ]

There is potentially a huge diversity on the amount of heat dissipated compared to nameplate ratings on the equipment, but unless the client has a similar existing space with similar equipment doing similar tasks, and you can measure the actual peak loads, just go with connected load (plus lights etc).

The type of computing being done has a great influence on the percent utilization of the servers. Some operate in a batch processing mode where they are nearly idle for most of the day, then running flat out for a couple hours. IT people can give you this info, but you have to design for the peak no matter what. If you press the IT people, they will give you a room temperature band that is acceptable, make full use of it!

 

[IRstuff]

I wouldn't get too complacent about the overdesign. Server rooms have a tendency to get stuffed with ever increasing numbers of boxes. What is overdesign today might just be barely adequate tomorrow.

TTFN

FAQ731-376