"For example, the solar panels are useless when it's cloudy, and wind turbines are only active when there's wind to spin the blades."
Finally someone's explained that to me /s
Power distribution specialist Eaton has kicked off a pilot project in Ireland that uses data centre UPS systems to help power the grid in periods of peak demand, to avoid grid-level outages. The owners of UPS systems are compensated for the effort – being paid as much as €50,000 per year for every MW of UPS capacity they …
You could harness the output of Trump.
I was actually thinking earlier today that he could single-handedly solve all of the worlds energy problems!
First, there is the thermal differential caused by his super-chilled heart and the vast amounts of hot air he produces.
There is the noxious gases that emanate from him - surely something that vile must have a fair amount of volatile material in it!
Speaking of volatile, there is his temper - that too should have considerable amounts of usable energy behind it.
And we really must do something about the piles of bovine excrement that pours like a wide river from his mouth.
Not only could he solve the world energy problems, he'd also solve world hunger! (though personally I'd rather starve...)
The Register should start a campaign for IT to become Greener.
The amount of electricity wasted in IT is criminal! IT is helping to kill the planet!
For example, in thirty years working in the industry I've never once seen site failover implemented (in a DR event) - not once. (And if you do insist on having it - how many times a year do you test it - not the six month UPS test - a full site failover?!) . All that gear in the redundant site whizzing away and never ever will it be used! (And if it is there is a 99% chance it will not work due to lack of testing and poor design)
Depends on the type of fail over event you are stating, we do once a year for DC core services being failed over and tested (Always my job :( and always a weekend, but at least its only once a year), someone else has the joy of going to the DR office and having machines built. We do have a last resort rush to PC world/Sainsbury's/Tesco and say give us every computer you have in stock, but obviously the bean counters don't like that one being tested.
"if it is there is a 99% chance it will not work due to lack of testing and poor design"
I thought of that too when the PR mentioned airports and stuff exporting power to the grid.
Here's an unplanned "test" the BAA prepared earlier (Gatwick, 2013) but I'm sure there are others more recent.
I saw a backup generator kick in once at the start of a power cut.
Driving through the gates of a data centre when the lights went out and there was an enormous bang and cloud of black smoke.
We thought the DC had been bombed (troubled times) but it was the backup generator firing up.
Impressive is an understatement.
It's almost as though National Grid's Short Term Operating Reserve never happened (via the Balancing Mechanism or not), and that aggregators/operators such as Flexitricity never existed, and subsidy harvesters such as Green Frog never turned redundant industrial sites into places to park containerised diesel generators for near-instant response to short notice supply shortages where there's a few quid to be made.
The early way is Essex: https://www.theade.co.uk/case-studies/demand-response/colchester-hospital
uses the hospital's own standby generators to feed into the grid (or take some load off the grid), if there is a short term need for less grid demand and more supply margin.
And then there was V2G too.
And *long* before all that there was (and still is) the grid-scale pumped storage system at Dinorwig. Zero output to a GW or so output in 10 seonds, but admittedly takes a bit longer to build, and probably cost a bit more, than a 200kW generator in a container.
It's almost as though National Grid's Short Term Operating Reserve never happened ...
Mind you, those all operate on the timescales of minutes. Near me, they recently finished a "grid scale" battery tank that could (in theory) supply 43MW for 30 minutes - or a little over 20MWHr. Can't find figures for it's cost as all the ones I can find lump it in with other stuff for a total price tag of (form memory) £1.5 billion. I vaguely recall from a talk I went to that it was in the £hundreds of millions range - making it 10, 20, or more £/WHr of storage. For truly grid level storage to avoid firing up peak lopping sets, you will need more than a few tens of MW capacity.
But what it is designed to be used for is second by second balancing (frequency control) - gust of wind increases output at a large windfarm for a few seconds, suck some power into the batteries; cloud passes over a large solar farm, let a bit out of the batteries.
But it's eye wateringly expensive to do grid scale battery storage.
As you say, the likes of Dinowig are good for the 10s+ range, diesel etc are good for the "couple of minutes"+ range. It would be interesting to see how they plan to do the communications needed for lots and lots of "small" UPS systems doing peak lopping at the single-second scale of control.
Allowing the power company to draw from your UPS seems to be like renting out the water from your fire extinguishers to alleviate shortages during a drought. All very lovely (I can just imagine the beancounters wanting to sign up to "maximise the return on the hardware investment") until a rare confluence of events tips you into the mire and you find your piles* are empty.
There's a reason UPS are installed, and I don't think it is because of the infalibility of the power supply networks.
*To clarify for the younger reader, an old term for a battery, not the result of management action once it all goes wrong.
You are correct Sir! The symptom is the energy company cannot supply enough power to customers is a good warning that the UPS will be needed for your own purposes. For example when there is a high demand for power, there is a higher chance the power will go out. A company with a UPS which plugs into this contract has management betting that there will be small general power reductions which can be resolved by the UPS but not small general power reductions followed by a large power reductions where the UPS is needed for the data centre.
I am of the opinion that UPSes cause as many outages as they solve but the hope is that the outages are at scheduled days and times.
Of course it won't help if the whole grid is out of capacity. But it should work for the much more common scenario of local shortfalls that require voltage regulation or very short term bridge capacity while power is shunted from another part of the grid. So more of a using your fire extinguishers to put out the fire in your neighbour’s garage kind of thing.
It could even help the whole grid situation if the shortfall is small and it's enough to keep things going for a few minutes until peaker plants can be brought up.
The cost does seem silly high though, at that price it would be more economical to build dedicated grid backup batteries.
Scenario 1 - There is not enough power available on the grid leading to a brown/black out. Your UPS discharges stored power until grid power returns to normal keeping your servers up and running.
Scenario 2 - There is not enough power available on the grid leading to a brown/black out. Your UPS discharges stored power until grid power returns to normal keeping your servers up and running; also, you get some money.
I mean, I know which one *I* would pick.
Some UPSes are for power filtering and covering until the generator starts up. That would leave extra capacity for reducing grid load.
The problem is when the UPS is drained down to the minimum reserves but the grid is still overloaded. Turning the mains power back on would not only resume normal load, but add the load of recharging the UPS batteries. The other option would be starting the generator. I don't know about the generators in Ireland, but the ones in Silicon Valley run about as clean as a tire fire.
"Some UPSes are for power filtering and covering until the generator starts up."
Yes. But the battery capacity is usually sized for this mode of operation. Just enough capacity to make it from the outage onset through the diesel startup. Or baring that, just enough time to command the racked servers to execute an orderly shutdown.
Been doing that (or similar) in the states for years. Large industrial business I am associated with has a massive diesel generator, UPS, and automatic transfer switch. They buy electricity very cheap, and lots of it. However, up to 5 times a year, with a 10 minute notice, they have to off the grid if the power company says so. Failing to drop off sees a 500x rate increase for that time period. So they hit the power button on the generator, it comes up to speed, 60 seconds later the automatic transfer switch dumps the load. No one notices, employees keep working, computers & lights stay on, no big deal. Besides, the generator needed exercise anyway.
Ireland's state owned grid operator, Eirgrid has fairly big plans to use battery storage on a large scale. There are some interesting proposals floating around. The aim, like most countries, is to get to very high levels of renewable energy and that can only be done with storage and large scale pumped storage isn't really all that feasible without high mountains and a willingness to flood vallies.
However, I've never quite understood by countries like Ireland, or the UK for that matter, actively chase data centre investments. Neither of them have very big green electricity production capabilities at present beyond wind power and data centres don't exactly generate big employment or taxable income flows. The biggest chunk of Irish power is currently coming from natural gas (domestically produced and imported). They'd surely be best located in places with tons of hydro power like Norway etc at least until wind and other sources have caught up.
It just seems odd encouraging big power hungry data centres into grids that don't have all that much spare generation capacity and in the Irish case at least, a surging economy and serious difficulties meeting agreed CO2 targets.
Britian seems to be in an even more precarious position with the aging and very low capacity nuclear power fleet and the Tories having largely failed to deliver on any replacement. Projects have been falling through while everyone's distracted by the chaos of Brexit.
However, at least if data centres are helping to balance peaks and troughs in the grid it reduces their impact somewhat and makes them somewhat better energy citizens.
Ultimately using batteries like that also means the whole network benefits from being able to use more wind power although we're probably going to need to see micro battery storage in most commercial premises and even homes to achieve that as well as large scale dedicated storage systems.
Ireland has always used a fair bit of CHP (combined heat and power) buy back from big commercial users, waste to energy, biogas etc etc ... I guess this is really just an extension to the same concept. Smart grids bring in a lot of potential flexible small energy sources to help deal with peaks.
Generally, compared to the US anyway brownouts are unheard of here. The grid and local distribution is very well managed and invested in. I've never really had any concerns about needing to protect equipment from power fluctuations in Ireland, but they were a huge issue in some parts of the US - surge arresters are fairly essential kit. Could also be to so with lots of urban air con inductive loads Vs air con being fairly unnecessary in Ireland beyond large commercial buildings.
I wonder how the Irish and British grids will cope with the rise of large numbers of heat pumps though. Could become interesting!
I understand that the ring at Fermilab uses a huge amount of electricity to charge their superconducting magnets (10% of the available grid electricity). They negotiate, cycle by cycle, with the power company to level the load. This gets them a cheaper rate.
The superconducting magnets do require energy to keep them running despite being superconductive.
This story seems a bit wide of the mark. Ireland operates a single electricity market with NI (oops! Brexit!). Power providers bid their *lowest* price to participate. The market purchases electricity at the price set by the highest contributor (it pays to low-bid). When demand is excessive or bids are high, the market requests the biggest users to drop out. Data centres switch to backup gennies. Less to do with brownouts and more to do with controlling the purchase price. DCs have been doing this for years. I suspect the "innovation" here is Eaton feeds into the grid at a high cost per MW, should the market price go above their bid.
Biting the hand that feeds IT © 1998–2020