[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: DATACENTER: AC and colo sizing question.
Anyone who has ever removed the thermostat from their car would realize the
fallacy of this concept. In order for a cooling medium to work, it must have
"dwell time". IOW, enough time in contact with the heat source, such that,
it can efficiently absorb the heat from the heat source. This time-value
depends on the thermal conductance properties of the medium. In this case,
it is air and this is why HVAC units do not run continuously. This is basic
physics, related marginally to mechanical engineering <grin>.
A smaller room would reduce the heat-sink properties of the resultant
air-mass considerably. It creates too tight a coupling between the HVAC
units and the heat-sources. I would not want that utility bill. Also, should
your HVAC units fail, the room temp would rise, almost instantly, well
beyond tolerable levels.
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of David Labuskes
> Sent: Thursday, February 10, 2000 8:13 AM
> To: [email protected]
> Subject: RE: DATACENTER: AC and colo sizing question.
>
>
> I've been working on an idea...
>
> My question is this. Everything about data center environmental issues is
> centered around ambient temperature, etc. I recognize that these
> should be
> 75 degrees, 55 humidity, etc. But if you put your hand next to the air
> exhaust of a server or even just touch one...you know that they are hotter
> than 75 degrees.
>
> My thoughts are to reduce the size of the "room" dramatically and
> create an
> environment of dust free rapidly moving air that will allow for
> the exchange
> of heat at a rapid pace. To make it all work though, I need to know what
> the maximum operating temperature is for this equipment. Not the "room"
> temperature. But more realistically, the motherboard or circuit board
> operating limitations.
>
> I apologize if the question doesn't make sense. Please feel free to reply
> privately or via the list.
>
> Thanks in advance for any and all information.
>
> David Labuskes, RCDD/LAN, CSI
>
> -----Original Message-----
> From: Julian, Dennis R. [mailto:[email protected]]
> Sent: Thursday, February 10, 2000 9:46 AM
> To: foo; [email protected]
> Subject: RE: DATACENTER: AC and colo sizing question.
>
>
> I agree, an educated client is a wonderful thing. In regards to peak
> cooling if you have over cooling what you lose is the ability to control
> humidity. It may feel cold and damp. The other concern is if the
> A/C units
> have compressors rather than a central chilled water source the increased
> number of starts resulting from short cycling may shorten the unit's life
> and increase the maintenance required.
>
> But being over cooled is better than over heated. One way to
> deal with this
> is through multiple units and redundancy. With multiple units
> you can stage
> in by manually or automatically turning on additional units as the load
> increases (or as the temperature rises). This can be done by using
> redundant units that also provide the ability to maintain cooling when
> unit(s) are down for maintenance or problems.
>
> Dennis R. Julian, P.E., RCDD
> Technical Associate
> Critical Facilities Design Group
> van Zelm Heywood & Shadford, Inc
> Mechanical and Electrical Engineers
> 29 South Main Street West Hartford, CT 06107-2420
> Phone (860) 521-4329 x373 Fax (860) 521-5620
> [email protected]
> www.vanzelm.com
>
>
> -----Original Message-----
> From: Roeland M.J. Meyer [mailto:[email protected]]
> Sent: Wednesday, February 09, 2000 10:40 PM
> To: foo; [email protected]
> Subject: RE: DATACENTER: AC and colo sizing question.
>
>
> > From: [email protected]
> > [mailto:[email protected]]On Behalf Of foo
> > Sent: Wednesday, February 09, 2000 6:54 PM
> >
> > Ken Woods wrote:
> > > On Wed, 9 Feb 2000, foo wrote:
> > >
> > > 8.34 A @ 120 = 1 KVA
> > > Total KVA * 2728 = BTU
> > > 12000 BTU per ton of air.
> > >
> > > Also add whatever heat load you have from lights, people,
> walls, support
> > > equipment (UPS and air unit, etc)
> >
> > My question is not so much how to calculate cooling for a known
> heat/power
> > load, but rather how to deal with transient increases in load while
> > equipment is being staged or housed temporarily or whatever as well as
> > gradual increases over time (as equipment is added
> permamnently). As I was
> > led to believe that engineering cooling for peak load would be very
> > inefficient and/or ineffective during times of non-peak load,
> I'm curious
> > as to what your suggestions may be to alleviate this problem.
>
> It is not near as inefficient as losing equipment due to
> heat-death. This is
> the inevitable result of NOT planning for peaks. Generally, add
> up the power
> usage numbers on all the equipment you are installing, allow for 5 warm
> bodies, add 30% for additional equipment arriving on a 100 degree day, and
> you should have enough. I like 70F +/- 5F.
>
> > > If I may, I suggest you hire an engineer that has a basic
> grasp on what
> > > you're trying to do, and isn't trying to guess. There are
> several on the
> > > list that would be glad to work for you.
> >
> > I'm not at all adverse to hiring someone to help us with this,
> however I'd
> > like to understand it a bit better before going that route. In my
> > experience plenty of facilities haven't gotten this quite right
> even with
> > "experts" on board (I've been a customer or visitor in quite a few
> > Internet Data Centers with serious environmental problems), so simply
> > signing away the responsibility to some third party without first
> > educating myself a bit doesnt make me entirely comfortable.
>
> You'll find, on investigation, that recommendations were usually
> under-followed due to desire to reduce costs. The cost difference
> between a
> 3 ton AC unit and a 5 ton unit is amazing. Many managers figure that they
> can "tough it out" during the few hot days, or decide to "policy exclude"
> additional equipment. More often, that manager gets another
> offer, elewhere,
> and their successor doesn't have their notes and they have to take the hit
> on their watch.
>