Conditioners' efficiency

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.
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...
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

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.

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]
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Andrew Gabriel wrote:

Is this surge higher than the nameplate specs?
>, but this reduces considerably once they've

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
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

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.

It's almost entirely the compressor in the outside unit. The indoor fan draws perhaps a few hundred watts.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

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
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
abu wrote:

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.

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).

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.

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.

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.

--
Paul Hovnanian snipped-for-privacy@hovnanian.com
-----------------------------------------------------------------------
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Paul Hovnanian P.E. wrote:

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! ??
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

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.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
James Sweet wrote:

Correct. It also won't influence the unit's efficiency. That is; the number of watts needed to move a BTU.
--
Paul Hovnanian mailto: snipped-for-privacy@Hovnanian.com
------------------------------------------------------------------
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
James Sweet wrote:

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
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

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.

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.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
James Sweet wrote:

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
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
abu wrote:

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.

Consult qualified HVAC and electrical designers. You appear to be making a much larger problem out of this than it really is.

--
Paul Hovnanian mailto: snipped-for-privacy@Hovnanian.com
------------------------------------------------------------------
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Paul Hovnanian P.E. wrote:

15000 KVA is the power of the computers, not of the A/C
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

You have 15 megawatts of computers?!
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
James Sweet wrote:

WHOPS ROTFL!!! We have 15 kVA of computers not 15000 sorry.... :-)
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
| Paul Hovnanian P.E. wrote: |>> 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!
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 |
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.