Hi Nick. Initila thought re Fan. I don't think a fan would work. All they do is blow the air around while actually adding to the internal temperature via the energy burnt off to produce the air movement.

Ah sorry, teaching grampa to suck eggs again

Tony, exactly what I'd do in my own house (if ever that were too warm!) but it'd be some sort of Heath Robinson arrangement. For this I need some sort of off-the-shelf or easily-contract-able solution. It occurs to me that I need to ask them whether the AC unit runs 24/7 or only office hrs. I think any fan would have to run likewise.

Some servers actually do some useful work at night - in fact probably most do except those purely serving internal desktops. But there's also a cargo-cult view that computers fail sooner if they're switched on and off - there's some truth to that in some cases, but not enough to justify the power usage nowadays, IMHO.

I'd be careful about venting air from the server room into offices; you might create a noise problem. You might also need to think about filtering new air going into the server room if you pump it through at large volume.

Some sort of night cooling system might work. How much power is the server room drawing?

Thanks all,

You have highlighted (?highlit?) the fact that I need more specific info. GreenPaddy, I think that, while your assumptions need comparison with the actual, you are probably right that the 'air dump' method won't work. I don't know enough about air-con - try never to use it. How 'clean' would the exhaust air from the AC unit be? Djh, you refer to the noise possibility - at the moment the AC machine thrashes away pumping air to the roof void. Fancies fly to some sort of plenum chamber/noise interrupter box with a low wattage 'out' fan taking air to a cooler office - absolutely no idea whether that would work! Any thoughts?

Nick

DJH..."cargo cult"...now there is a good name for a band!

http://www.myspace.com/cargocultmusic

Someone beat you to it.....

Nick,

In terms of cheapness, I wonder about an air-to-air HP? Could that be a way of removing the excess heat without the disruption of ducting? Yes, I know there's still the refrigerant pipes, but there may be some smaller holes to break out? Does anyone know at what temp they usually blow out heat?

Nick

The missing link in most cooling-the-equipment thinking, from server rooms through commercial kitchens, printing presses and manufacturing processes generally, is the assumption that it has to be done, as hitherto, by diluting or cooling the room air.

That's a recipe for paying good money expensive high-grade energy - electricity, or fossil burnt at high-temp - then deliberately degrading it by dilution or heat-pumping to the point where it's just useless low-grade 'waste' heat to be got rid of - and usually paying for more high-grade electricity all over again, to power the the wastage process.

In a server room, first use servers designed to minimise power consumption - CPUs that use only a fraction of the power are on the way, and the average (cheap) PSU is ridiculously inefficient. Then get away from cooling such heat-emitters by blasts of room-air provided by lots of cheap little inefficient fans. The heat-emitters in a computer run at quite high temp, and their cooling system should be arranged to output the smallest-volume coolant flow at the highest possible temp - so that the heat is captured at source and doesn't get to heat the room-air at all. Having captured it, at as high-grade temp as possible, then it can be easily taken away and re-used for something useful.

Same could be applied to e.g. commercial kitchens - instead of gas hobs blasting their gaseous heat into the room-air, induction hobs heat the pans direct, even through a pan-insulation jacket that minimises loss from the heated pan/contents. Cooking vapours taken away into ducting plumbed to the pan lids. There's a nice bit of system-design for someone! Lots of very high-grade low-volume ducted coolant to play with, and minimised room-heat and escaped fumes to dispose of. Contrast with the conventional ventilated-hood method, which relies on high air velocity to exceed the velocity of the volatile fumes' ability to back-flow, driven by their own partial vapour pressure, against the air flow. The result is massive dilution of the valuable high-grade flame-temp heat output of the hobs. The diluted flow-temp of an extract hood is useless in every way - and a significant part of that useless waste heat is from the electricity bought to power the fans whose job is to waste all that originally-valuable heat!

Posted By: SteamyTeaIt is not just a CPU that generates heat and has to be cooled. Hard drives, video cards, PSU's, Mother and any Daughter boards and RAM all generate heat that needs to be dissipated

Yes, all those. But "has to be cooled", "needs to be dissipated" - that's the mindset that regards the heat as a nuisance to be got out of the IT box into the room, where it becomes the building's problem, rather than a resource to be captured. Effective capture can happen in any building, Brunel or not doesn't matter.

Thanks Paul

So you had any experience in retro fitting this kind of stuff.

While on the topic, who makes a very low powered PC, I did see one that claims to only need 8W and was not much bigger than a couple of fag packets. Can't find it now.
I would like one to plug my sensors into for some remote sensing.

Nick

Posted By: fostertomlazy profligate heat-pumping then not necessary. You know that, Paul.

If you put the evaporators in the server room you can get COPs of 6 or 7 - it's much easier to move the heat around as a liquid rather than a gas. So it's hardly profligate - and all the "waste" energy that's used to run the heatpumps is ultimately used where it's needed anyway.

Any large server room will have water cooling on the equipment anyway which allows the heat to be moved without a heatpump - though later down the chain it may need a heatpump to make it usable again. This is all standard practice these days anyway, especially for LEED certified offices where the building is designed as a system and all heat sources and sinks are considered. Same thing in commercial laundries too (since you mentioned other sources of waste heat earlier) - heatpumps are used to recover the latent heat from the driers which is then reused in the washing water.

Paul in Montreal.

Nick, getting back to trying to solve the problem you have today, I'd like to steal/repackage a thought from Fostertom, (around using the heat before it's diluted). I'm just thinking out loud here so bear with me.

When I've seen these racks before, they are a bit like a tall filing cabinet skeleton, which takes each of the computery bits like 150mm drawers (don't know what you've got). Would it be possible to have some sort of a shroud built around the rack system (back, side, top), and then connect the extract leg of you MVHR unit to this, drawing air directly across the rack units. Because you're concentrating the heat take off whilst it hotest, you might get extract air at 40oC??? Pull in fresh air from outside at (well -15oC today) but lets say 10oC ; a 30oC deltaT? Send the warmed air to your cold people's offices.

Normally MVHR can't do much space warming (I think) but that's cause you 're moving air that is not much different in temp to that already inside the house. But you might be moving air in the high 30's oC??

There are probably lots erronious assumptions above, and probably overheat risks to the hardware, should the air draw fail, but hidden in the ramblings might be something that someone else could pick up and repackage again in another way - sort of "e-pass the parcel" - well Fostertom started it.

GreenPaddy.

GreenPaddy

I like the idea, but how about just taking the heat from the top of the room and allowing the cooler, fresh air in at the base of the room. Then the whole room becomes a huge heat exchanger/mixer. Odours could be a problem as well as noise transmission.

Tom

I understand what you are saying about using more high grade energy (electrical) to shift and concentrate low grade energy (dissipated heat), but energy is energy, and in this situation you would have either a lot of low grade or or a little higher grade. You would still have the same number of Joules to play with regardless of how it is transported around the building. Energy storage and control is the key to successful energy management.

Posted By: SteamyTeayou would have either a lot of low grade or or a little higher grade. You would still have the same number of Joules to play with

That needs crystal-clarification, otherwise a big source of confusion. As you say, "same number of Joules", therefore "a lot of low grade or or a little higher grade" isn't true. The amount of energy is the same, whether low- or high-grade, but if allowed to deteriorate (so easily) to low grade it's useless, no longer available 'to be played with'; if consciously conserved in highest possible grade it's useful.

A small volume/flow-rate of thermofluid (air, water, refrigerant etc) at high temp is useful; the same amount of heat in a diluted larger volume/flow-rate at low temp is useless.

The latter arises because it's easier, more compact, more techie and lazier to deploy large delta-t temp differences and/or high volume/flow-rates to facilitate the heat transfer into the thermofluid (e.g. to cool a server room with blasts of refrigerated air). It takes much bigger, engineered, old-fashioned heat transfer interfaces, and design-intention, to effect that heat transfer at minimum delta-t and at minimum thermofluid volume/flow-rate, so as to get a small volume/flow-rate of thermofluid at high temp, which can then be used for something else, without heat pumping it it to higher grade.

It's not good enough, to regard waste-heat streams as merely an advantageous source for heat pumping. Raising the heat pump's COP from 3 to 5 amounts to buying yet more electricity whose sole purpose is to waste only 20% instead of 33% of the energy in the waste-heat stream. That is no substitute for conserving that waste-heat stream at high enough grade to be 100% useful without heat pumping.