Data Center World: Flywheel UPS Demonstration

[bigman55434]

I think you would have a tough time convincing some datacenter owners to place their reliance on 15 seconds of power. If you could beef up that flywheel and get at least 1 minute of power, that gives you a little better window to recover via generator.

[lcabosa]

bateries for cars, flywheels an fuel cells for energy storage, how dificult is to get that ???

[TOGGI3]

Sometimes, that 5 minutes of backup makes all the difference in customer power failure. Flywheel depends entirely on the system automatically kicking over to generator power. Anything that could possibly make that fail and require manual intervention makes the flywheel system useless.

[cdzieciolandmisie]

a 300kw diesel generator that's been sitting for a year or two takes about a minute of cranking to start up. 15 seconds of power just won't do the trick

[Waynebarkr]

@cdzieciolandmisie Not true--a well maintained genset ( if you have one installed in a data center it better be) will start up and take the load in 5 to 8 seconds. Gensets in parallel will start up and parallel in less than 20 seconds.

[TOGGI3]

@Waynebarkr Yeah but what happens when something happens the switchgear doesn't like that requires manual intervention, say abnormal utility power, and it doesn't switch? Or a generator fails to come up? These things can and do happen.

[CurtHowland]

@cdzieciolandmisie "that's been sitting for a year or two"

Someone has been cutting corners again.

A mission critical generator should be run tested once a week.

[dubyamoney]

@cdzieciolandmisie What!!!!! are you hand cranking it old timer??? the whole point in having the gen set is to ensure continuity, you only have a small 300kw unit, I've just installed a 1.5 Mw unit, I would be shot if it took 15 sec to start!!! PS make sure you have a maintenance regime in place and test it weekly, oh yeah!!! don't forget the fuel!!!

[Digalog]

apc is not very happy with these vid i assume

[mcsemel]

And yes, we use straight AC-to-AC from the grid to the data center. There is no double-conversion as found in battery-based UPS systems (AC-to-DC-to-AC) so we don't get the heat loss and loss of energy. And there is no 'whiplash effect' in a flywheel-based UPS.

[Digalog]

wow thanx for the explanation

[mcsemel]

We have one of these units powering our data center. It has been in place since 2002 with a 750 kW generator. We have not had an issue with startup and have not had to supplement it with batteries. It has taken over 250 power events from the grid since it was installed with no effect on the protected equipment. Our UPS has a GenStart unit so the flywheel starts the generator if the UPS detects a loss of grid power.

The $1,000,000 savings coupled with its 7+ year track record speaks for itself.

[dubyamoney]

@mcsemel 250 power events??

[hdinfo1]

Every engineer I know who installed a flywheel system has had to go back and supplement it with batteries once presented with the real world. I'd say that eliminates conjecture.

[marek35]

did he say 200 kW on this one wheel? that is a lot of energy

[xtranet1]

250kW

[scottylans]

55,000 rpm baby - if that thing shatters, hot jesus

[shawnlhood]

7,700rpm; not 55,000rpm

[antronx007]

Not 200KW of energy, its 250KW of POWER for 15 seconds, which equals to 1.04KWh of recoverable energy in that flywheel. His 98% efficiency claim is vague. Is it AC in to AC out efficiency? That's very unlikely. Also, he did not mention friction losses over time. This thing may actually use more energy to stand by rather than battery based UPS. With lithium nano-phosphate batteries, you can get 250KW of power for 2 MINUTES using only about 15 KWh battery pack with much better system efficiency.

[marek35]

Thanks, now I understand, 250kW for 15 sec. is 1,041 kWh. That is not much energy :(

[Sirrom0206]

This fly wheel is pretty small! The ones we had in my sister plant were the size of VW beetles. My plant had battery backup UPS. Yes, 0-batts equal less floor space. but, what happens if the genset don't auto-start initially? Where is the 20-minute battery buffer for T-shooting the genset? This is great if everything works right. But in an emergency nothing works right, thats why its called an emergency outage.

Just in time back up power.

I must be getting old because I don't trust this.

[TehMG]

Yeah I was wondering about this myself. Sometimes generators don't start on the first try if they have been sitting unused for a while. In the event that it takes a few tries to get the genset going, is 15 seconds really enough?

[happyguy82]

What frictional losses? Modern flywheels have a spin-down time measured in Years! NASA has many such flywheels, suspended by magnetic bearings in a vacuum. Look up NASA's G2 flywheel.

[Digalog]

can you explain that last part

[MatthewLWspace]

15 seconds is not a reliable amount of time for a generator to start, it becomes many times more impractical if you involve any kind of generator paralleling gear. Flywheels should really only be used in addition to battery based systems.

[roidroid]

>"15 seconds is not a reliable amount of time for a generator to start"

then use 2 of them for 30 seconds, or 4 of them for a minute, whatever you need.

Your complaint could be used exactly the same for batteries: "oh one battery only gives you X amount of power" - well that's why we use more than one battery in a system.

And just the same: you can use more than one flywheel in a system.

[MatthewLWspace]

Clearly the best solution is to limit your thresholds to the second instead of those extending to ten fold the typical threshold of batteries. Proper research validates my points above your conjecture.

[roidroid]

i don't know what you just said

[krazykizza]

Put me down for 4 my good man.