Data Center, Colocation, Cloud Computing, Storage, Dedicated Servers Forums

- Data Center Design, Development, Building Systems and Operations (http://www.datacentertalk.com/forum/forumdisplay.php?f=37)

- - cooling tonnage versus ambient temp (http://www.datacentertalk.com/forum/showthread.php?t=7261)

cooling tonnage versus ambient temp

Does anyone have links to information about relative energy efficiency of split-system datacenter cooling at different ambient air temperatures? For example, the difference between cost of keeping a datacenter at 65F versus 75F?

Perform calcs as below to find your loads. Dropping sensible load to 65 will lower your humidity which may require humidification. It depends on how you want to operate. Bottom line, the more energy you use to condition your space, the more expensive your costs. The upside to keeping your space at 65 is you have created a heatsink during cooling failure. May only be 30 minutes or so, but may be enough time to correct a minor problem.

Below taken and edited from RSES SAM Manual..... (www.rses.org)

Is = Np x Fs + Lw x 3.41 + Lf x 4.25 + NM x HP x 3393 + As

where:

Is=sensible internal heat gain, Btu/hr

Np=number of occupants

Fs=occupancy heat gain, sensible, Btu/hr

Lw=electric lights, incandescent, watts

LF=electric lights, fluorescent, watts

NM = number of electric motors

HP = motor HP

As=appliance heat, sensible, Btu/hr

IL=NP×FL+AL

where:

IL=latent internal heat gain, Btu/hr

FL=occupancy heat gain, latent, Btu/hr

AL=appliance heat gain, latent, Btu/hr

Use actual or assume a fluorescent lighting load of 1.5watts per sq ft of floor space.

Use actual or assume (For FS and FL) 250 and 200 Btu/hr per person respectively.

Quote:

Originally Posted by jcchiefeng (Post 13378)

The upside to keeping your space at 65 is you have created a heatsink during cooling failure. May only be 30 minutes or so, but may be enough time to correct a minor problem.

Thanks for the formula, John. But I should point out that at higher densities, depending on air as a thermal ballast during data center cooling outages doesn't scale, since racks just start to re-breathe their exhausts. I saw an interesting CFD simulation showing that in an environment built out to 400W/sq ft, a rack will overheat in 10 seconds.

Ken

Yes, you're right if the fans die....I was obviously not clear....my apologies

Actually, in the simulation it didn't much matter if the CRAC fans die or not -- the difference was only a few seconds. The effect happens at the rack level. Here's a whitepaper on it (registration required :mad: ): http://www.upsite.com/TUIpages/downl...Cooling_WP.pdf

Ken

Thanks for the info...it will help in my presentations....JC