Water, sunshine or pig s***: How will YOU power your data centre? So you've decided to clean up in Big Data. You're going to have the server farms of all server farms, run the cloud for the world's megacorporations and of course there's this bloke in Silicon Roundabout willing to give you the money. What's your first decision then? It should be to go and steal an idea from another industry …
I did te calculations on GSHP and frankly, it was marginal with oil prices for heating. The real killer was the need to totally redesign the whole house central heating system to deal; with lower departure flows. That plus the heat pump and electrical work, was going to cost around 15 years of oil purchases.
And whilst in a freezing power cut, its possible to rig a generator to run the central heating, a power cut would mean end of all heating with GSHP.
Whether they scale well enough to deal with amount of heat put out by a big datacentre though, I've no idea.
No, except in very cold regions or special circumstances (eg there's a big aquifer not far under the datacenter). You can only economically shove so much heat into the nearby groundwater (which is what a GSHP does) in a given amount of time; once you start heating the groundwater, you lose efficiency until conduction and convection move that heat away. That's first-year thermodynamics.
That's why GSHPs often work well for private homes in uncrowded neighborhoods (though as others have pointed out, unless energy is expensive in your area, you're not likely to save much money), but aren't generally suitable for big commercial installations.
Isn't the underlying problem that you're "dumping" the heat? All that electricty goes in, is used to spin the electronic wheels, and most of it comes out as lower-grade energy in the form of heat. Which is thrown away.
Never mind just using natural cooling to discard the heat cheaply, why not find a way to use it, instead of dumping it? Put the data centre beside a district heating plant? Beside a hospital?
This post has been deleted by its author
"(on a serious note, why is this a law? how do we know for certain? surely it is still just a theory supported by evidence..)"
Have a read of this: http://www.evolution.mbdojo.com/theory.html
Quick summary:
Laws and theories are very different things in this context. Theories don't get magically promoted to be laws; scientific theories remain theories forever. Heliocentricity is just a theory, as is phlogiston.
A law is a description of an observed phenomenon. A theory is an explanation of that phenomenon. A theory must necessarily be consistent with a law. Laws may not explain their whole domain; Newton's law of gravity works fine for two point masses for example, and issues like the predicted orbit of Mercury didn't invalidate the law, just limited its application.
The problem with energy-reuse schemes is that the needs of the parties never completely coincide. The hospital or district heating scheme has no need for heat in the height of summer, but you need to dump your waste energy 24/7. Equally you need the freedom to take your server farm offline occasionally for maintenance, without worrying about freezing the local residents. Both parties need to invest in backup plant which will spend most of its life idle.
Actually there is one ploy that works.
In certain places. Take Scandinavi. Hot summers. Cold winters. SO you bury your heat exchnger in the ground under a car park, and in summer, you pump aircon heat into the ground. In winter, you use a heat pump in the aircon to pump the heat out of the ground and back into the building.
I id d some calculations,. A storage heater consisting of nothing more than a house sized insulated cellar full of near boiling hot water, would heat my house for days, if not weeks. Enabling me to use off peak electricity at near oil prices.
If the desired output is large amounts of low graded heat for space heating, simple thermal stores are effective and cheap. And the bigger they are, the less heat they lose. area goes up as square of linear, volume as cube. Het content is as volume, etc. Massive blocks of concrete, or ponds of water, keep heat very very well. Even just an area of land, and soil with no insulation at all, will be reasonably efficient.
Supermarkets do ths as well. The refrigeration heat pumps provide the hot air curtains in winter. Only in summer are they actually dumping heat into the environment
BUT they all have big car parks. Why not bury a heat exchanger under those?
Its an idea that makes sense when electricity costs rise beyond a certain point. My guess , we have reached that point.
As I said above - ground source heat pumps are only good for heating - not for cooling a data center or much else for that matter unless the ground stays cooler than the outside air.
Especially a data center where you have no need for heating anytime of year - unless there is a prolonged powercut in winter above the arctic circle - in which case your ground source heating won't work either. You could use the heat to warm up the ground for the local civilians communical heating system... but then to pump the heat into the already warm ground would cost you even more than pumping the heat into the air.
Maybe the should do like some airport in japan - scoop up all the snow in winter and use if for cooling during the warmer months. But I guess you need a rather large carpark or runway to make that at all useful.
"ground source heat pumps are only good for heating - not for cooling a data center or much else for that matter unless the ground stays cooler than the outside air."
Heat moves both ways. It's a conduction thingie. (Convection helps, at a personal level, but we're not exactly typoeing about human comfort, now are we?)
The trick is in the plumbing ... Know your system, know your ground. It's not rocket science.
A Bathonian's really most unlikely say anything nice about Stroud now, are they. Nor Radstock in the opposite direction to be honest.
Fun story about Radstock (just to show that I'm generally prejudiced, not specifically). Brother's in the catering trade. He once turned down an opportunity to take over a pub in Radstock on the grounds that the area was vastly too dangerous to work in.
He's spent the last decade in Helmland and Kabul running logistics chains. There's danger and then there's real danger in his reckoning and Radstock doesn't come out of it well.
In the Chiltern woods you'll find lumps of slag from primitive (Iron Age and medieval) iron smelting, yet the nearest iron ore is many miles away. Just as Tim points out, you need 20 tons of wood to make the charcoal to smelt a ton of iron, so it was much easier to carry the iron ore to the wood than vice versa.
This post has been deleted by its author
"Has anyone tried doing waste heat recovery on all the hot air that comes out of a data centre? "
Search on "low grade heat recovery". Many have tried, few have found much success, simply because the energy in low temperature waste heat is not that great.
If you can think of a use for large volumes of lukewarm water, or warm air then you should let the electricity industry know, because they have a similar problem of huge volumes of low grade waste heat at power stations.
Yes has been done at least once in London town, somewhere in Docklands. Otherwise heat the local swimming pool or the district energy system with your data centre which have been trailed in Europe.
I wrote a MSc on waste heat recovery and yes recovering energy from low grade heat sources sucks - as stated numerous times above.
" large volumes of lukewarm water "... once nice use of otherwise waste heat from a local geothermal plant is a prawn farm;
http://www.hukaprawnpark.co.nz/history
There are other agricultural options such as hothouses growing orchids or similar as well.
If somebody really is looking for a data centre somewhere with a bulk supply of 'leccy on tap, I would suggest Manapouri - underground, vast amounts of cold water flowing through and 800MW on tap would power a lot of ARM server blades. That power is currently mostly going into a smelter, but that is looking like it could be shut down, leading to an excellent lair/data centre for evil genius types.
( http://en.m.wikipedia.org/wiki/Manapouri_Hydroelectric_Power_Station)
"And you also need to be able to cool the room after your machines have turned much of the generated energy into heat."
"Much" of the energy is turned into heat? What happens to the rest of it? Hint: for all practical purposes, _all_ of the energy is turned into heat. There is a utterly negligible amount that leaves the room via network cables, but for the purposes of specing an HVAC system, all energy is turned into heat.
Why generate the heat in the first place?
1) Scrap the Xeons and replace by I7 - reduces power/performance by 30%
2) Run shop at 12V rather than 240V and consolidate all the inefficient power supplies, applying diversity to optimise power supply - this could give you an extra 20% reduction or more.
3) You can now cut the cooling by half - saving more electricity.
You could always make the ops sweat by raising the temperature from chilly to say 30 deg C
If only IT type companies accepted remote workers, living in the North and asking companies based in and around London if remote working is available is a lesson in futility.
However I do know quite a few people in other businesses that are accepted as remote workers, for some reason IT companies can't keep up with the pace of IT or do not trust their workers
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
