Hacker News Reader

Comment by noodlesUK | Hacker News Reader

PG&E has been doing blackouts and brownouts more than basically any other electric company I’ve ever experienced anywhere in the world since I was a little kid. These blackouts may be bigger, but if you’re an institution the size of a university, you should have significant disaster preparedness infrastructure. I mean US universities run police forces! I’m sure they could find a way of designing redundant power into their systems. They might even save money in the long run, and wouldn’t need to worry about campus hospitals or public safety going down in an outage (I’m sure some of that is already taken care of, but why not go all the way).

View on HN

Many universities have "cogeneration" plants that produce electricity, steam for heating, and also hot water. However, they usually don't generate enough electricity to power the whole campus, so the administration chooses who gets it and who doesn't. If these sorts of outages are going to become more common, then the management is going to have to re-evaluate the cost of providing backup power for all labs vs the cost of lost research.

The university of California at Berkeley has ample cogeneration capacity but they are tied to PG&E grid and so PG&E can still turn them off.

Thanks for the clarification; I was wondering what had happened to their cogeneration plant.

I remember some time ago when the cogeneration plant failed and took out power in one building where many biological samples were stored at -80.

Wow, that's annoying, too bad that they wouldn't have a way to disconnect. IIRC, Princeton was able to operate some buildings off of the cogen plant during Hurricane Sandy, so I guess they had things engineered differently.

All grid tie inverters shut off if power in the lines are down for safety reasons. You would need something special to switch over.

Yes of course, I just would have assumed the cost of that is pretty minor compared to the cost of the (apparently) 26 megawatt cogeneration plant

I don’t know — just imagine how many places the mains power is tied into all of the buildings. Each of those places would also need a mains cutoff, and they would need to all be thrown/active at the same time before the cogen plant could be activated independently. At a place the size of UC Berkeley, that would be a pretty big cost (and logistical nightmare).

If the system wasn’t designed for the switchover from the beginning, I can see how it would be considered impossible/not practical to do. Especially when this type of “blackout” was thought to be rare.

I’m sure the prior assumptions would be reconsidered today.

> just imagine how many places the mains power is tied into all of the buildings

Not many? As in, I don't think many buildings would have two separate power feeds. I'd expect a small number of very large connections from the cogen plant and the outside word into local distribution network(s), and then everything is single-path from there on.

But isn't that just because of legislation? As far as I can tell the system could just as easily throw a switch, test to make sure the switch was thrown, and then safely power locally. I know in California this is law, but I have never heard of a very good reason why it is technically difficult or inherently more dangerous to "air gap" your local power at the building rather than at the battery/power source.

Agreed. When I lived in Wisconsin the power would go out frequently during storms. Immediately I'd hear some neighbor's natural gas generators kick in and then a few minutes later I'd hear a few neighbors fire up their gasoline generators.

In both cases, they would flip over their home (or a sub panel) to the alternate power source (and disconnect it from the upstream grid). I looked into these systems and the natural gas ones could do this automatically using a kit available at Home Depot, and the gas generators required a manual switch, but were still very simple.

I looked into Solar in California and when I found out that most people can't use their systems during blackouts I asked about these generator systems that switch over and they basically said they had not seen equipment that could it and they didn't believe it was allowed anyway.

Solar is different from a generator, in that a generator is already producing AC, and as long as it produces a reasonable frequency and waveform you're good. Solar produces DC, which you then need to convert to AC, and the inverters that are typically installed are not able to do their own waveform shaping, depending on the grid to do it for them. It's possible to get inverters that handle this correctly, but they're more expensive, so typical home solar installs don't use them.

Or is it a horse and cart problem? Since there is legislation against it, there is not a demand, and prices have stayed high to sell them as "off grid" tech only? Most anybody would spend $500 extra on a 30k system for guaranteed power during blackouts.

There is no legislation against it where I am (Massachusetts), and there is still very limited demand for it because of the pricing.

> Most anybody would spend $500 extra on a 30k system for guaranteed power during blackouts.

To be clear, to get "guaranteed power during blackouts" out of solar, you need both a more expensive inverter ($1k or so last I checked) and batteries ($5-15k last I checked). If you're OK power only when it's sunny _and_ a blackoutm you might be able to do just the more expensive inverter, maybe. I'm not actually sure whether that work without a battery sink for your unused current.

> I'm not actually sure whether that work without a battery sink for your unused current.

What happens to unused current now when the panels are disconnected during a blackout?

There isn't any. That's not how electric current works. If you leave a disconnected solar panel out in the sunlight it doesn't explode or whatever.

Nothing, since they are disconnected, so there's a voltage across the gap but not enough to spark. And you don't care precisely what the voltage is, because there's nothing happening with it.

But if you don't disconnect them, now you are maintaining a voltage across whatever your load is, but that voltage depends, for solar panels, on the current being pulled from them, as I understand. And you need to maintain the voltage in some nominal range. Maybe it all works out, but I'd have to do a lot of digging to be sure...

Any inverter is expensive. I am currently pricing out a small system and just found out today the battery ready inverters don't cost more. In fact I am going without batteries due to a niche use case, but the inverter will be battery ready just because it (SolarEdge) is the best deal overall for the system. You are right about battery cost, but just having one that can get you (frugally) through the night would be a world different than just being shafted.

As to battery requirement, I can't see how attaching a wire to a panel would cause it to burn up. The wire would charge and then it would be the same thing as if the panel was disconnected with the same amount of latent charge in the panel.

Inverters are only used for DC to AC conversion, cogeneration sets are conventional prime mover setups that output 3-phase AC directly. The problem likely has to do with islanded grid stability, without the grid you have to take care of power factor correction and grid voltage regulation yourself, this can get pretty dicey if you don't have direct control or knowledge of the type of loads you'll have.

In an islanded mode of operation you also need to match load and generation to maintain system frequency, which is tough to do with cogen and non-controlled loads.

Basically, it can work, but the whole system is going to need to be engineered for it from the outset.

I recall some time back, the power generated with cogeneration needed main power with the right frequency to condition(?) it. The cogenerator couldn’t run during a blackout.

For a blackout, the diesel generators kicked in (and could completely power the platform).

Power plants need to be connected to other sources so their AC frequencies can be in phase. If the multiple AC waves were out of phase, you’d cause all sorts of issues. It wouldn’t surprise me if the cogen plant was designed to only work in concert with external power, but not in place of it.

Emergency power would be expected to work independently of mains power.

True for inverter based generation (most renewables like wind and solar), but generally not for rotating machines like most Co-gen plants. Some rotating machines (steam turbines generally) won’t have governors for frequency control, but usually in a CHP plant they would have a linked combustion turbine that does. Rotating machines that are paralleled with the rest of the grid are by default in sync with all of the other rotating machines if there is not some sort of fault situation at the plant or on the grid. If you try to parallel a generator with the grid that is out of sync (frequency or phase), the grid will always win and pull the generator into sync (potentially destructively for the generator).

The most likely limitation of a typical power plant operating disconnected from the grid would be if it could “black start” or not. Cogen plants would normally have this capability. Essentially the problem is whether you have enough power available from backup generation to bring all of the plant auxiliary systems online and get the generators initially started and producing power. Once started you set the governor to maintain 60.0Hz (if in North America) and begin slowly bringing on load so you can ramp the generation to match it.

The easiest solution to a future blackout would be to match the load in your island area to the output of your plant and then disconnect from the rest of the grid before or at the time the blackout occurs. This way you have avoided the black start situation completely. The blackout detection, disconnection, and load balancing steps can be automated to occur nearly instantly if the system is designed well.

The main issue for Berkeley is that the cogeneration plant doesn’t seem to be sized to accommodate their islanded load. This is especially problematic if you lose a generator while islanded because then you will have instant frequency collapse if you were already on the edge of your plant’s capabilities.

There are different types of generators: some supplement, and others replace, the grid.

Cogeneration is designed to produce heat (in chillier climates) and the process makes electricity as well while doing so, because why not?

* https://en.wikipedia.org/wiki/Cogeneration

Emergency / standby generators do nothing most of the time, and only kick in when needed:

* https://en.wikipedia.org/wiki/Standby_generator

* https://en.wikipedia.org/wiki/Emergency_power_system

I work for a significant tech company in the Valley and we are in the PG&E grid, but we have diesel backup generators that we immediately fired up when power was cut to a few of our buildings. “Tied” to the grid doesn’t preclude backup power. As long as we can get fuel trucks to our buildings, we can run indefinitely.

If you think that lack of disaster preparedness is bad, apparently many dialysis clinics in California don't have the generation capacity to continue operating during a power outage like this one. No big deal, I mean they're only keeping people alive after all...

Iowa State University has it's own power plant. Surprised that isn't more common.

University of Cincinnati, Michigan State, lots of large universities have that.

Seemed like it should be fairly common. Saw in a different comment that maybe there is some issue with being able to switch over to only being powered by that plant.

Seems odd though, since one of the things I thought was always touted about our municipal power plant growing up was that they could power the town if external power got cut off (But didn't normally because our generators were more expensive to run)

It's a law in California. One that I am sure PG&E lobbied for.

U.C. Berkeley does have one too.

Not really: I was recently in Juticalpa, Honduras (it's a major city, like SF) and we did not have light for three days. Similar experience in Rio Dulce, Guatemala. In the third world, doing science is very difficult, because the infrastructure is just not there.

But the cost of providing full backup for 5 days has got to be pretty significant. Likely it means diesel generators the size of locomotives, sourcing fuel in what may also be a fuel shortage crunch, maintaining and testing this plant.

EDIT: I screwed this all up and it's wrong. I tried. I'm sorry. Leaving it up so someone else can give it a go. And there's already a bunch of comments.

Washington University uses about 376,394 kWh/year. [1] That's about 7,841.54 kWh/week. Washington pays about 13.1 cents per kWh for power [2]. Running on the electric grid alone this would cost $598.59 per week. This isn't what they pay though, because they offset much of this usage with renewable energy like solar panels.

That would take a 50kw generator to power. Assuming they don't power all systems during a blackout they could probably get away with a 25kw generator. For argument, let's assume a 50kw generator full load would probably use about 4.0 gallons of diesel fuel per hour. [3]

At $3.64/gallon for diesel [4], at 4.0gallons/day, for one week the university's fuel cost would be $2,446.08.

To recap, the university would spend $2,446.08 to generate electricity they could purchase for $598.59 from the electric company during a 7 day blackout at full capacity.

[1] https://www.electricchoice.com/blog/25-of-the-most-energy-ef...

[2] https://www.bls.gov/regions/mid-atlantic/news-release/averag...

[3] https://www.dieselserviceandsupply.com/Diesel_Fuel_Consumpti...

[4] https://www.eia.gov/petroleum/gasdiesel/

How can a university draw only ~50kW? That’s not even one rack of computers. I think you missed a large factor somewhere.

UCB has a 26MW gas turbine.

That's one and a half industrial combi ovens (each takes 480V 3 phase 38kw power). Those numbers seem so low.

Cause op didn't read citation 1 where it says that the University saved that much power by switching to better light bulbs

Yeah, I'm sorry. I edited it with a disclaimer as soon as I realized.

You're welcome to continue retracing my steps to correct me though.

I was going to say something similar, maybe this is the core critical components that have to say running for life safety.

Otherwise I routinely see single racks pulling 15-25kWh albeit vastly overloaded. But that’s the norm now.

Yeah, from the link, the quoted number is actually the amount Wash U saved by changing light bulbs.

> the school has completely replaced their lights with more energy efficient, low wattage bulbs that save an average of 376,394 kW hours per year in energy

About UCB: “The cogeneration plant uses a natural gas-fired GE LM2500 turbine generator to produce 25 MW of electrical power and 4 MWth of steam.”

[2013] https://www.powerengineeringint.com/2013/10/14/university-of...

Here in CA the price is about 0.29c kWh, so that’s $2,275. Pretty comparable to running a generator.

“0.29c kWh”

WHAT?

Here in France I pay 0.138c (€, thus 0.15c in $) for 100% renewable elec. That's like twice as much in CA... for what I guess is not environment-friendly by a long shot?

I'd really have thought with the USA fracking their soil like it's apocalypse tomorrow, and pulling out of any carbon commitment on top of that, at the very least it would yield considerable short-term domestic advantages, i.e. cheap elec. I'd have thought you guys paid like 5 or 10c kWh. Next thing you're gonna tell me you pay more than $1.2/L ($4.6/gal) of diesel fuel.

Actually just checked, diesel fuel is $3/gal, regular petrol slightly less in general (~$2.5) but more in CA.

Still baffled at the cost of electricity.

There are parts of the US where electricity is cheap, and a few parts of California where unleaded premium fuel is at or near $4.99 per gallon. The pricing doesn't make much sense to me.

CA requires a unique gasoline mix, I think that’s responsible for a lot of the price diff overall.

California has a different gasoline mix with the seasons for emissions reasons. And every time we switch between winter and summer mix, the few refineries here always seem to have mysterious maintenance emergencies. Not sure why they always need to shut down for a few days right then, but it sure makes for a supply crunch and a price bump.

> That's like twice as much in CA... for what I guess is not environment-friendly by a long shot?

You'd be wrong. CA has one of the cleanest energy mixes in the country, arguably among the cleanest in the world for it's size:

https://www.eia.gov/state/?sid=CA#tabs-4

I'm so happy to learn that! I wish I could edit the parent post.

Way to go, CA. Really. I imagine it's not exactly easy given the 'context'.

Have a look at the data here: (click on the place you want to look at)

https://www.electricitymap.org/

What an awesome website. Thanks a lot

France has a massive amount of nuclear power with subsidized capital costs.

Surprinsingly enough, the French nuclear electricity costs about as much as hydrocarbons. It seems the "almost free electricity!" slogan of the 1960's was less than accurate in its predictions.

Lots of valid reasons though, I don't know that we've been particularly bad at this. Nuclear fission is just not the insanely cheap producer economic theory would suggest on paper, at least not in all configs/countries.

There would be no reason to charge the end customer less than the hydrocarbon electricity, as long as the demand is there.

It could be 0.05c/kWh less and still be an economic solution.

In New Zealand I pay ~29c US/kWh for 80% renewable generation during the day. 11pm-7am it runs about US10c/kWh. Handy for charging the car. I live ~500km from the main generation source as well, until a larger local geothermal plant comes online.

CA has much higher electricity rates than most other parts of the US. For example, in Texas electricity costs about $0.11/kWh; Washington it costs about $0.09/kWh; and the average in the US is ~$0.13/kWh.

Probably the disparity is in part because of how likely it is in CA for power lines to burn down entire towns, thanks to a dry climate with many large forests adapted for fires; CA is so naturally fire-prone that some native Californian tree species, like redwoods, require large forest fires as part of their reproductive cycle. So more line maintenance is required (although PG&E has probably not done enough...).

I see. Well, the national average seems more in line with what we'd infer from the macro.

Thanks for explaining!

Redwoods do not require forest fires for reproduction, and no native species require large forest fires.

There are plenty of native plants that need fire to reproduce, but they do better with smaller fires that produce heat to open seeds but don't damage topsoils.

The large fires we have now are the result of climate change, long-term fire suppression, and unsustainable forest management. It's not normal for California, even though there are fire-dependent habitats here.

"Require" was too strong a word for redwoods, but FYI they are helped by fire when reproducing and "often require" it: https://en.m.wikipedia.org/wiki/Sequoia_sempervirens

I agree that CA is much more at risk of fire now than it naturally is, thanks to short-sighted human endeavor; I was only meaning to point out that it isn't correct to compare CA electrical rates to France as a measure of U.S. vs French rates, since CA's are so much higher than the U.S. average due to fire risk.

Thank you for writing that out!

I’m guessing the problem is in the capital cost of having a generator system installed. Getting procurement, university real estate/capital projects, a general contractor, and any deans/admin involved to build the system.

I have just assumed that all science buildings had backup generators by default. Movies may have tricked me. I think all hospitals are required to though, which makes sense why the researcher moved his metabolic stuff to UCSF

As user jumpingmice pointed out the obvious error in your numbers, the figure you've cited from [1] as the yearly power used was the yearly power saved by adopting efficient light bulbs:

> Throughout the rest of the campus, the school has completely replaced their lights with more energy efficient, low wattage bulbs that save an average of 376,394 kW hours per year in energy. Their overall lighting plan has saved more than 20.6 million kilowatt hours in total.

GIGO

I think we got trolled :)

Nah, I was giving it an honest go. Just not diligent enough I guess.

Can't delete it now though!

I think some of your citations are off.

The 376,394 kWh/year is the savings they got from going to low wattage bulbs.

"Throughout the rest of the campus, the school has completely replaced their lights with more energy efficient, low wattage bulbs that save an average of 376,394 kW hours per year in energy" [1]

And your prices are from the Washington-Arlington-Alexandria municipal area, not the St. Louis area in which Washington University St. Louis is located.

[1] https://www.electricchoice.com/blog/25-of-the-most-energy-ef... #7 on the list

[2] https://www.bls.gov/regions/mid-atlantic/news-release/averag...

Reply on news.ycombinator.com