Conditioners' efficiency
abu
we need to cool down a server room.
It is known that we need 3500 BTU every KW of computers (plus some more
for the room form factor, ok)
I was reading that conditioners nowdays have an EER rating of 10, that
is they draw 10000 BTUs for every KW of consumed electricity. Hence, I
computed that we would need 350W to cool down 1KW of computers.
However, our electrician tells us that this is not true, because the
power drawn by conditioners is in facts much higher than what is stated
on the box. He says that a fully running 12000 BTU conditioner actually
draws 3 KW, not 1.2 KW like it is written on the box.
If this is true, then we would need 875W of conditioners in order to
cool down 1KW of computers, instead of the 1.2 KW declared on the
conditioner specs.
This would be a problem for us, because we would then need to place new
electric cabling in order to bring more current to the server room.
Do you confirm that conditioners are so much more inefficient than what
is stated?
Thank you.
Oh, another question: where are these conditioner watts dissipated? I
hope that is on the external fan unit, not on the in-room AC split,
otherwise we would need to also cool down those watts, and the needed
conditioning power would be immensely superior...
Andrew Gabriel
abu <abu...@abeabo-net.it> writes:
> Hi all,
> we need to cool down a server room.
> It is known that we need 3500 BTU every KW of computers (plus some more
> for the room form factor, ok)
>
> I was reading that conditioners nowdays have an EER rating of 10, that
> is they draw 10000 BTUs for every KW of consumed electricity. Hence, I
> computed that we would need 350W to cool down 1KW of computers.
>
> However, our electrician tells us that this is not true, because the
> power drawn by conditioners is in facts much higher than what is stated
> on the box. He says that a fully running 12000 BTU conditioner actually
> draws 3 KW, not 1.2 KW like it is written on the box.
>
> If this is true, then we would need 875W of conditioners in order to
> cool down 1KW of computers, instead of the 1.2 KW declared on the
> conditioner specs.
>
> This would be a problem for us, because we would then need to place new
> electric cabling in order to bring more current to the server room.
>
> Do you confirm that conditioners are so much more inefficient than what
> is stated?
I think your electrician is getting mixed up. He will have to size
cables, breakers, supply, etc for higher current as aircon takes a
surge when starting, but this reduces considerably once they've
started up.
It's not a good idea to run your aircon from the same supply as the
servers though. This startup surge can generate quite a dip in the
voltage at the end of a supply cable. Depends on your supply impedance
at point of use.
> Oh, another question: where are these conditioner watts dissipated? I
> hope that is on the external fan unit, not on the in-room AC split,
> otherwise we would need to also cool down those watts, and the needed
> conditioning power would be immensely superior...
The external heat exchanger gets rid of both the heat pumped out of
the room, and the energy consumed by the aircon.
--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]
James Sweet
"abu" <abu...@abeabo-net.it> wrote in message
news:4815dd7f$0$87943$892e...@auth.newsreader.octanews.com...
> Hi all,
> we need to cool down a server room.
> It is known that we need 3500 BTU every KW of computers (plus some more
> for the room form factor, ok)
>
> I was reading that conditioners nowdays have an EER rating of 10, that is
> they draw 10000 BTUs for every KW of consumed electricity. Hence, I
> computed that we would need 350W to cool down 1KW of computers.
>
> However, our electrician tells us that this is not true, because the power
> drawn by conditioners is in facts much higher than what is stated on the
> box. He says that a fully running 12000 BTU conditioner actually draws 3
> KW, not 1.2 KW like it is written on the box.
>
> If this is true, then we would need 875W of conditioners in order to cool
> down 1KW of computers, instead of the 1.2 KW declared on the conditioner
> specs.
>
> This would be a problem for us, because we would then need to place new
> electric cabling in order to bring more current to the server room.
>
> Do you confirm that conditioners are so much more inefficient than what is
> stated?
>
> Thank you.
>
>
The quoted efficiency is accurate under one specific set of conditions. Amp
draw of the compressor is pretty much directly related to condenser
temperature. On a hot day outside, or if the coils are blocked by debris,
the head pressure and amp draw will be much higher than on a cool day. The
only one I've actually measured is the 36,000 BTU unit at my house, I've
found it draws between 10A and 14A at 240V, yours is probably somewhat lower
efficiency, but even the 1.2KW sounds a bit high and is probably measured on
a hot day.
Your electrician likely has a poor understanding here, refrigeration isn't
something a lot of them understand well. He's probably taking other factors
into account to choose the wire size and breaker capacity rather than the
actual power draw.
>
> Oh, another question: where are these conditioner watts dissipated? I hope
> that is on the external fan unit, not on the in-room AC split, otherwise
> we would need to also cool down those watts, and the needed conditioning
> power would be immensely superior...
It's almost entirely the compressor in the outside unit. The indoor fan
draws perhaps a few hundred watts.
daestrom
"abu" <abu...@abeabo-net.it> wrote in message
news:4815dd7f$0$87943$892e...@auth.newsreader.octanews.com...
The heat of the A/C is discharged by the external fan unit, not back into
the room.
But look again at the EER rating. Most A/C units are rated in SEER
(SEASONAL Energy Efficiency Ratio). In the US, the 'seasonal' rating is
based on a long drawn out calculation that tries to mimic the entire season.
Some days with a mild outdoor temperature and better a EER, and other days
with killer outdoor temperatures and a lower EER. Then they crunch a lot of
numbers to figure out 'typical' performance over a 'typical' season.
Not that much use in a specific application, but good for comparing similar
units in the store.
You know the amount of heat you need to reject based on the load in the
room. If this is a large server room, don't forget the lighting load (count
the number of 40 watt flourescent tubes in the ceiling to get an idea) as
well as any peripherals like a bank of monitors or such. Is this room the
only room in the building to be air-conditioned? If so, then there will be
a lot more heat load between it and other rooms. If the whole building is
already air-conditioned and you just want to keep this room from
overheating, then you're on the right track. Multiply by a 'fudge factor'
to account for inaccuracies, growth, humans in the room, etc... (say, an
extra 20%)
Then you go find the A/C unit for *that* amount of heat load. Look at the
A/C wiring requirements for that unit and have the electrician run a
separate circuit for it. Don't plug it into the same circuit as your
computers (nothing good can come from that).
Good Luck
daestrom
Paul Hovnanian P.E.
>
> Hi all,
> we need to cool down a server room.
> It is known that we need 3500 BTU every KW of computers (plus some more
> for the room form factor, ok)
>
> I was reading that conditioners nowdays have an EER rating of 10, that
> is they draw 10000 BTUs for every KW of consumed electricity. Hence, I
> computed that we would need 350W to cool down 1KW of computers.
This assumes that they are running at 100% duty cycle and doesn't allow
for additional capacity for other factors. It gives you a good idea of
the long term energy consumption of the a/c system, but isn't as useful
for sizing the units.
> However, our electrician tells us that this is not true, because the
> power drawn by conditioners is in facts much higher than what is stated
> on the box. He says that a fully running 12000 BTU conditioner actually
> draws 3 KW, not 1.2 KW like it is written on the box.
Your electrician is incorrect in the narrow context of his statement.
All equipment draws no more than what it says 'on the box' (on the
nameplate, actually) per code requirements. It may draw more on startup,
but panel, breaker and circuit sizing are based on nameplate ratings as
dictated by code (I'm assuming the NEC for your jurisdiction).
> If this is true, then we would need 875W of conditioners in order to
> cool down 1KW of computers, instead of the 1.2 KW declared on the
> conditioner specs.
What you need to do is to consult with a qualified a/c designer. Server
rooms are not residences and cooling equipment requirements are
different.
> This would be a problem for us, because we would then need to place new
> electric cabling in order to bring more current to the server room.
If you have to install more capacity, it's got to be done right.
However, I doubt that the capacity will have to be provided to the
server room itself. Most heat pump installations are either outdoors or
in separate equipment rooms. That's where the new circuits will be run,
as specified by a qualified designer.
> Do you confirm that conditioners are so much more inefficient than what
> is stated?
It depends on the application of the equipment. The number you read may
have been for residential occupancies and calculated with the
appropriate load factors. Server rooms are different animals.
> Thank you.
>
> Oh, another question: where are these conditioner watts dissipated? I
> hope that is on the external fan unit, not on the in-room AC split,
> otherwise we would need to also cool down those watts, and the needed
> conditioning power would be immensely superior...
--
Paul Hovnanian pa...@hovnanian.com
-----------------------------------------------------------------------
Procrastinators: The leaders for tomorrow.
abu
>> However, our electrician tells us that this is not true, because the
>> power drawn by conditioners is in facts much higher than what is stated
>> on the box. He says that a fully running 12000 BTU conditioner actually
>> draws 3 KW, not 1.2 KW like it is written on the box.
>
> Your electrician is incorrect in the narrow context of his statement.
> All equipment draws no more than what it says 'on the box' (on the
> nameplate, actually) per code requirements. It may draw more on startup,
> but panel, breaker and circuit sizing are based on nameplate ratings as
> dictated by code (I'm assuming the NEC for your jurisdiction).
I don't understand what you say here. If the A/C might draw more than
specified on the nameplate when it starts up, and we size the breaker
for the nameplate specs, at the AC startup the breaker will trigger and
break the circuit!
??
abu
> I think your electrician is getting mixed up. He will have to size
> cables, breakers, supply, etc for higher current as aircon takes a
> surge when starting
Is this surge higher than the nameplate specs?
>, but this reduces considerably once they've
> started up.
>
> It's not a good idea to run your aircon from the same supply as the
> servers though. This startup surge can generate quite a dip in the
> voltage at the end of a supply cable. Depends on your supply impedance
> at point of use.
This is interesting. All of you have said this.
We certainly would have this problem, as the cable reaching the server
room is like 50 meters long, and both the A/C and servers are attached
at the end of it.
I am a CS engineer, not electric, hence I have some understanding of
electrical engineering but not a deep one, and I cannot really
understant how this "dip" happens.
If we size the cables for the worst case, that is, maximum A/C
(air-conditioner) current draw + maximum servers current draw, even
during the A/C engine start the voltage available to the servers should
not fall below the specs of the power supplies. I understand that there
would be "interferences" in the sinusoidal shape of the alternated
current but, who cares: the power supplies of the computers AFAIK work
like a rectifier + charge pump and not like induction transformers, so,
sinusoidal interferences are irrelevant in this type of design, they
should practically have no effect at the output of the power supplies...?
Thanks a lot
James Sweet
"abu" <abu...@abeabo-net.it> wrote in message
news:4818342d$0$43460$892e...@auth.newsreader.octanews.com...
Breakers do not trip instantly.
The spec you're looking for is LRA, Locked Rotor Amps. A typical 3 ton
residential AC unit may draw 12A operating, but the LRA can be over 100A.
That's why the lights in the rest of the house will dim momentarily whenever
the unit starts up. This current surge is not long enough to cause the
breaker to trip.
Michael A. Terrell
Computer power supplies draw current at the peaks of the AC line.
The startup current of the compressor motors lowers the peaks for
several cycles, until the motor is pretty well up to speed. Us a
voltmeter to monitor the AC line while an A/C kicks on and you will see
this. Even better, borrow a scope and look at how much it dips. The
meter is weighted, so it shows trends, while a scope shows the details.
During those dips, the filter capacitors in the power supply are not
fully charged. Depending on the quality and age of the supply, it may
cause problems, and the chances go up as the capacitors age. Being
servers, they are probably on 24/7/365.24 which ages the parts even
faster than in a desktop used during working hours. If you aren't sure
of what is going on, contact your electric utility and see if they have
a recording monitor to log changes in the line voltage over a 24 hour
period. Also, keep in mind that the problems will be the worst on the
hottest days. I have had to straighten out CATV head A/C problems, as
well as in Tv station's control rooms.
You haven't said how many servers, or how they will be housed. If
they are in enclosed racks, a raised computer floor will reduce the
required cooling, by pumping cold air into the bottoms of the racks, and
pulling the head out of the tops without releasing the heat into the
room.
--
http://improve-usenet.org/index.html
Use any search engine other than Google till they stop polluting USENET
with porn and junk commercial SPAM
If you have broadband, your ISP may have a NNTP news server included in
your account: http://www.usenettools.net/ISP.htm
phil-new...@ipal.net
The typical circuit breaker has two trip elements. One is thermal, which
requires reaching a certain temperature to trip the breaker. The thermal
element very closely mimics the heat buildup in the wiring it protects.
A very brief 100A starting current will heat things up more than a 50A one
of the same duration, but this won't be all that much for just a second or
two. The other trip element is a magnetic one. It is supposed to catch
short circuit faults more quickly. It is adjusted to a point expected to
be well above any motor starting current in the targeted type of usage.
Many industrial circuit breakers, where very large motors are involved,
have an adjustment for this right on the front of the breaker.
If you have a circuit with a 20A breaker, and then start drawing 30A on that
circuit, it will not trip the breaker immediately. You may have a few minutes
before it trips. For higher current levels, that time is shorter. But it
should be long enough to get most motors started just fine.
If you have an air conditioner that is rated to draw 1.2KW, but draws 3.0KW
all the time, then either it is defective, or the manufacturer lied about the
rating. But I would not be surprised if it draws 6.0KW for a second to get
started. For an air conditioner that large, I'd get one for 240V, not 120V,
and wire the circuit accordingly. That would reduce the starting current in
half, resulting in half the voltage drop, which would be split between the
two 120V "sides".
--
|WARNING: Due to extreme spam, I no longer see any articles originating from |
| Google Groups. If you want your postings to be seen by more readers |
| you will need to find a different place to post on Usenet. |
| Phil Howard KA9WGN (email for humans: first name in lower case at ipal.net) |
Paul Hovnanian P.E.
Correct. It also won't influence the unit's efficiency. That is; the
number of watts needed to move a BTU.
--
Paul Hovnanian mailto:Pa...@Hovnanian.com
------------------------------------------------------------------
Life would be so much easier if we could just look at the source code.
abu
> Breakers do not trip instantly.
>
> The spec you're looking for is LRA, Locked Rotor Amps. A typical 3 ton
> residential AC unit may draw 12A operating, but the LRA can be over 100A.
Thanks to all of you, also to Michael replying on the other sub-thread:
I think I understand the whole problem now!
Now, this problem is going to be really nasty for us because due to some
policies of the building that hosts us, it seems that we cannot bring
two separate power lines to our server room.
In addition, I think we cannot afford an UPS so big to protect all the
equipment. We accept to have the equipment go down if there is a
blackout (this is a computation cluster, we don't need 24/7 guaranteed
availability), however we would like to protect the equipment from the
damage that can be caused by the frequent brown-outs from the A/C. Is
there anything we can do that comes to your mind?
Two ideas come to my mind but I am not sure of the feasibility: would
you please comment?
1 - We could use a current limiter before the A/C. Does it exist? What's
the technical name? We need one that can be put before a 50,000 BTU air
conditioner (probably three-phase).
2 - We could use a buck-boost voltage stabilizer before the computers.
For high powers I think this costs 1/10 of what an UPS costs (we need
around 15000 KVA).
Thank you
James Sweet
> 1 - We could use a current limiter before the A/C. Does it exist? What's
> the technical name? We need one that can be put before a 50,000 BTU air
> conditioner (probably three-phase).
>
You can't do this, if you limit the current, the compressor motor will stall
at startup, this will cause the current draw to sit at the max limit until
the breaker trips or the thermal cutout in the motor opens.
> 2 - We could use a buck-boost voltage stabilizer before the computers. For
> high powers I think this costs 1/10 of what an UPS costs (we need around
> 15000 KVA).
>
That should certainly help. You could also pick up a smaller UPS to run the
most important equipment and leave some of it on just a stabilizer.
Paul Hovnanian P.E.
>
> James Sweet wrote:
> > Breakers do not trip instantly.
> >
> > The spec you're looking for is LRA, Locked Rotor Amps. A typical 3 ton
> > residential AC unit may draw 12A operating, but the LRA can be over 100A.
>
> Thanks to all of you, also to Michael replying on the other sub-thread:
> I think I understand the whole problem now!
>
> Now, this problem is going to be really nasty for us because due to some
> policies of the building that hosts us, it seems that we cannot bring
> two separate power lines to our server room.
You don't run the A/C power into the server room. The compressors and
condensers should be installed outdoors. There will be some fans in the
server room. But these draw a relatively small amount of power. It may
be possible to mount these outside of the server room itself.
> In addition, I think we cannot afford an UPS so big to protect all the
> equipment. We accept to have the equipment go down if there is a
> blackout (this is a computation cluster, we don't need 24/7 guaranteed
> availability), however we would like to protect the equipment from the
> damage that can be caused by the frequent brown-outs from the A/C. Is
> there anything we can do that comes to your mind?
>
> Two ideas come to my mind but I am not sure of the feasibility: would
> you please comment?
>
> 1 - We could use a current limiter before the A/C. Does it exist? What's
> the technical name? We need one that can be put before a 50,000 BTU air
> conditioner (probably three-phase).
>
> 2 - We could use a buck-boost voltage stabilizer before the computers.
> For high powers I think this costs 1/10 of what an UPS costs (we need
> around 15000 KVA).
Consult qualified HVAC and electrical designers. You appear to be making
a much larger problem out of this than it really is.
> Thank you
--
Paul Hovnanian mailto:Pa...@Hovnanian.com
------------------------------------------------------------------
Very funny, Scotty. Now beam down my pants!
abu
>> 2 - We could use a buck-boost voltage stabilizer before the computers.
>> For high powers I think this costs 1/10 of what an UPS costs (we need
>> around 15000 KVA).
>
> Consult qualified HVAC and electrical designers. You appear to be making
> a much larger problem out of this than it really is.
15000 KVA is the power of the computers, not of the A/C
James Sweet
"abu" <abu...@abeabo-net.it> wrote in message
news:481acba7$0$43445$892e...@auth.newsreader.octanews.com...
You have 15 megawatts of computers?!
abu
> You have 15 megawatts of computers?!
WHOPS ROTFL!!!
We have 15 kVA of computers not 15000
sorry.... :-)
abu
> You can't do this, if you limit the current, the compressor motor will stall
> at startup, this will cause the current draw to sit at the max limit until
> the breaker trips or the thermal cutout in the motor opens.
Hello again,
on this matter, I have just learned about the "inverter" technology:
this seems what we are looking for, doesn't it?
http://en.wikipedia.org/wiki/Inverter_(air_conditioning)
Do you confirm that buying this type of air conditioning unit should
avoid us the problem of the dip at startup (potentially damaging for the
servers) that everybody has mentioned?
This would be immensely cheaper for us than any other type of solution
like a current stabilizer or UPS...
Thank you