Be interesting to know what the fire supression system would have been if they had gone for a full li-ion system. Putting even a small li-ion fire out is non-trivial (says he having seen laptop battery fires).
The San Diego Supercomputer Center in the US is ditching its lead-acid uninterruptible power supply (UPS) batteries for more environmentally friendly rechargeables – though it's avoiding lithium-ion, and going with a new form of rechargeable alkaline. Currently, the SDSC relies on a generator and UPS to provide emergency …
"The only viable system would be oxygen starvation..."
Uh ... Maybe. As I understand it, when Li-ion batteries are warmed much above 400C they go into thermal runaway. I don't think the internal chemistry uses atmospheric Oxygen, so I'm not so sure that depriving the battery of Oxygen will shut the fire down. But it will (probably) keep the packaging and anything else in the vicinity from burning. So maybe Oxygen starvation will help ... or not.
If not, what might work is something with high thermal conductivity to get the temperature down and poor electrical conductivity to keep from shorting battery terminals on adjacent cells. Which is to say, not water.
Someone around here probably actually knows how to approach a large Li-Ion fire. My inclination would be to get the hell away from the thing and let someone else deal with it.
I don't think the internal chemistry uses atmospheric Oxygen, so I'm not so sure that depriving the battery of Oxygen will shut the fire down.
I believe you speak the truth. There was much talk of such things during the Samsung Galaxy Note debacle, and I'm sure I read that in the event of a fire even a bucket of water wouldn't help you...hence why airlines were so twitchy.
Li-ion battery fires produce their own oxygen unfortunately - only way to stop/reduce the oxygen production is to cool the fire. CO2 apparently helps reduce the intensity/duration of the fire for that reason. Doesn't put it out though.
Big fires? Dump a few tons of sand on it and wait for it to burn out. I can't remember how long it took the Aussies to put out the container fire on the SA "battery" but it wasn't quick.
"Someone around here probably actually knows how to approach a large Li-Ion fire. My inclination would be to get the hell away from the thing and let someone else deal with it."
If not trained and equipped, get the hell away and let someone else deal with it.
Otherwise, surround and drown. Li-ion batteries tend to provide their own oxygen, so nitogen, CO2, or similar systems don't help. The large installations also have layers of cells, so getting a class-D extinguishing agent to the seat of the fire isn't guarranteed. That leaves getting enough water on to keep it cool enough to prevent spreading the fire.
In storage facilities you do this with in-rack sprinklers. In production areas, if possible you submerse the entire battery ("dunk tank" style, or move to something like a large roll off refuse container.
Anon, because $dayjob doesn't like to publicly discuss how things can go bad.
I briefly worked a contract job rejigging the network in an EV battery plant. Starting with individual electrodes the entire packs were built up on site. The individual modules were loaded into huge conditioning chargers, rows and rows of them like the aisles in a supermarket but 20 feet high, and along each row there were dozens of nitrogen cylinders ready to dump their loads - there must have been hundreds or thousands of those cylinders in the plant. We were told that if an alarm went off, sprint for the exit or you'll asphyxiate.
Given that, as others have pointed out, lithium fires make their own oxygen, I can only assume they were there to stop any fires in the electrical systems from setting off the batteries? Or maybe just for show. With that and the emergency showers everywhere in case of an electrolyte spill (get naked and get in there in a hurry or the acid will eat into your bones and kill you) it was a jolly fun old place.
This is my understanding as well. They are very simple construction, the materials are basic and easily separated so recycling is cheap and easy.
This looks like some smart sales people using the "lead is bad" flag.
Pretty much any battery is "bad" if you break the case, not something that should occur in a UPS.
Most of the lead acid cases now are based on some sort of plastic that is also quite easily recycled as it is not mixed.
Stories that just parrot the PR release really frustrate me. OK, fine, bring it to my attention, but perhaps enrich it a little and do better than the original? It's very frustrating when you have quantities (5200 alkalines) compared to weights (20000lbs lead-acid). Also, energy storage amounts? The original press release you reference has this data verbatim for the new cells: Why not just include it (1MWh for 5200 cells, meaning ~192Wh/cell)? Otherwise just post the link to the press release :-D What about the relative energy density? That's NOT included in the press release - why not ask? That _would_ be interesting.
Giving usable and comparable numbers would be great but that would not make for a good story.
Good tech-stories, for the non-technical, are about feelings and diversion. Giving information that makes you feel good but not necessarily makes any sense under scrutiny. Please do not pop the feel-good bubble. Please keep consuming and do not ask any questions.
>"We can now offer hours instead of minutes of power to our UPS customers"
So what? I note the PR release doesn't give any real answer.
The normal is to have sufficient batteries to cover the generator startup lead-time.
Personally, as this datacenter is for academic/research usage, I see little point in having more capacity than is strictly necessary to cover minor supply fluxations and interruptions and can support a controlled shutdown for anything longer-lasting.
So I think this announcement is simply putting a positive spin on having to do something to comply with green legislation. I suspect also the new in-rack batteries can supply DC rather than AC that has to be rectified and thus resulting in shorter run time for a given UPS KVA.
- "Some portion of our datacenter is only on street power and goes out immediately when the grid fails," Kirkpatrick said.
What? Is this quote real?
My own datacenter days are long behind me and now I only have to keep eyeballs on a trio of old HP servers in a single rack but even they have a pair of basic UPSes to keep them running for a few minutes in case the power goes out. These people actually built a datacenter where bits of it go off 'immediately' if grid power fails?!?
It sounds like the municipality limited their use of generators, and when you are talking lead-acid cells in a datacenter, building structure becomes a real issue for the shear weight you are going to have to support (I've seen buildings that have brought in many tons of steal beams that had to be bolted directly to building structure in order to support the battery bank, so major re-construction of the building).
So they may not have had much choice in expansion of traditional lead-acid cells, or generators.
If you had to choose, its best to triage your load, vs. than just throwing too little to support it all...
But I'd agree with other comments, there seems to be more behind the story that could be reported.
> Because of environmental regulations in its home state of California, the SDSC has been unable to scale up the portion of its emergency power delivered by generators.
Sounds like an ideal application for redox-flow batteries. You don't have a tank of diesel fuel on-site, instead you have a tank of electrolyte. No air pollution on-site, no refuelling trucks and though not silent due to pumping electrolyte, runs much quieter than generators.
Biting the hand that feeds IT © 1998–2022