America's dramatic shift from gas power plants to batteries | Hacker News Reader

America's dramatic shift from gas power plants to batteries

theprogressplaybook.com

85 pointsepistasis73 comments

The article seems to mix storage capacity and instantaneous power. Lithium batteries tend to be able to discharge their stored energy in perhaps 1 to 4 hours (for instance the Tesla Megapack is between 2 and 3 hours to discharge).

Yes batteries are great at flattening the evening peak, when prices presumably spike upwards. That appears to be a good use-case, cycled once a day. Eyeballing the second graph there appears to be around 25 GWh of storage participating on that day, which makes is last equal in terms of energy delivered grouped by type (excepting coal which does not appear to have a line on the graph despite being on the key).

Most interesting is the large amount of imports.

When talking about grid storage, "capacity" refers to power, not to energy, because grid assets are almost universally identified primarily by their power. Changing this tradition for storage assets would be far more confusing than the correct terminology which is used in the article.

There's an entire industry, with highly trained professionals, advanced curriculums, etc. etc. etc. that run the grid. It's old technology, but it is still technology. They have chosen to measure battery storage assets primarily by the power capacity, because that's the most important metric for how grids are actually run in practice.

Every single time that outsiders are introduced to the grid, there's always some comments that assume that the grid people or the journalists somehow got the usage of MW and GW wrong, but it's actually how it's used in practice by the real engineers who run these things. I have had many many discussions on HN and other forums going over this, but I have not yet found the language to concisely convince people of this fact (and my comment is mostly addressed to other readers of the forum, not to you directly).

I understand that such a capacity metric might be useful for generation, and somewhat useful for storage. One could see the grid operators having a good grip on the utility of storage to smooth off tv pickup and evening peaks, in full knowledge that the energy stored will run out in a few hours.

However the linked article is on theprogressplaybook.com; the about page has the following blurb

  The Progress Playbook™ focuses on proven ideas for a better world. We cover the policies and projects that are succeeding in driving sustainable development, and that could be replicated elsewhere. This publication is intended to be a resource for policymakers, business leaders, civil society organisations, and ordinary citizens.

It is clear that the publication is not aimed specifically at grid insiders, so propagating the grid's argot to a more general audience is at the very least unhelpful.

Part of the problem is the source data. In one example the article mentions a battery is 680MW and has 4 hours of discharge. But most of the source data is a LBNL Interconnection Queue report that does not have that data available (the Council of Economic Advisors seems to be using this). I know the source report does not have this data because it is not published in the report and the lab does not have it (I’ve asked them).

Well, I've eyeballed the capacity of the storage in the system from the graph at around 25 to 30GWh. The CAISO website does allow downloading per-day CSV files of the 5-minute power on a per-day basis. There is also the question of behind-the-meter batteries which presumably do not show up (neither generated solar nor evening peak consumption saved).

I hardly ever see a popular science article about any kind of energy that gets units right or understands what they mean.

Usually they can manage when it’s a conventional power plant by parroting how many “homes” it could power. If it’s anything else forget it.

I expect batteries will replace peaker plants entirely. They're superb at delivering precise on-demand power in bursts, which is exactly the role those plants used to fill.

They also pair exceptionally well with solar power, because grid demand is highest in the few hours around sunset, and batteries can store the extra solar during the day to meter out over that peak period after the sun sets.

Yes and we've seen nothing yet on that front. It's impressive that at the current cost levels, batteries are already making a huge impact. But those cost levels are still high and there are much lower cost batteries on the way. Bigger, better, cheaper, etc. That's going to be the story with batteries for the foreseeable future.

Using the same batteries as in automotive applications, as is common in the storage market right now, doesn't make any long term sense. Those are optimized for things that simply don't matter in grid storage. Like size and weight. Energy density doesn't matter for grid storage. Cost per kwh is what matters. Decimating that cost is going to drive adoption further. And who says it will stop there? Using cheaper chemistries (e.g. sodium ion, metal oxide, redux, etc.) is going to open up much more of the market.

And while battery and renewables cost continues to come down; we're pretty much at the end of the road for gas plants. They've peaked technologically and they are about as efficient as they'll ever be. Worse, they might get more expensive to operate as they come under pressure to implement carbon capture. Also carbon taxes might increase the price of fuel for these plants. And of course the fluctuations in gas prices means operating gas plants becomes more risky.

There are already a lot of gas plants designed for 24x7 operation just sitting there and idling until the rare moment the operators can justify the cost of firing them up. Utilization rates for these plants are dropping. That's going to have an impact on related investments and the return on those. Anyone banking on 24x7 running operations is going to be in for a nasty surprise.

The article mentions a Californian plant from 2008 that has already been decommissioned. That's a great example of this. This is going to repeat across the market. Most gas plants operating today will be decommissioned long before their scheduled end of life and they won't be generating as much of a return on investment as was hoped for at the moment of their opening because of poor utilization and a shortened life span.

That's going to significantly affect the interest of investors in building new gas plants. That probably already is a huge factor.

Do you think it's realistically ever going to be possible to store months of energy in grid batteries? or is that beyond the scope of credible optimisations?

Serious question because articles like this [1] made me skeptical that renewable energy outputs could be transferred between seasons.

[1] https://www.technologyreview.com/2018/07/27/141282/the-25-tr...

Batteries are the wrong tool for that, storing large amounts of energy is the domain of pumped hydro.

While pumped hydro will definitely play a role in seasonal storage, I doubt it's going to be the main solution.

IMO, hydrogen and biogas are going to be playing dominant large roles in seasonal storage once the grids becomes so renewable-heavy that normal daily operation during normal weather can be covered by wind and solar.

Pumped hydro requires specific geography (lots of water, high and low points) and significant earthworks around it, which makes it unviable in a lot of places.

Pumped hydro is a bit problematic when all you have are tiny hills that freeze over.

When you’re talking even 3+m deep bodies of water they don’t freeze solid even in the arctic circle. A thick layer of ice on top still provides the same weight, so winter just doesn’t change anything.

There’s some countries where the geography doesn’t work out, but most midsized countries already depend on exchanging power with their neighbors simply on economic grounds.

It's doable given a large enough need. But I don't think there's much of a need to do this. People seem to blindly assume this but you have to challenge the assumptions they are making a bit.

Grid storage is not about long term storage at all but about demand shifting so that you can shift over production when demand is low to peaks (evenings typically) when demand is high. That's it. There are some setups with multi day capability but essentially most of the market is about 4 hours or less. That's what's killing the demand for gas plants. It's not long term storage. People build gas plants for base load that is supposedly needed. But then they end up being pushed into peaker-plant roles because the demand for base load just is a lot less than assumed.

The reality is of course that we actually have a lot of base load in the form of nuclear and gas plants. That's not going to go away overnight. We probably have more of it than we need. So as long as we have too much of that, demand for seasonal storage is not going to pick up. That might change over the course of the next decades. But it will be slow and probably cost driven.

Can we build huge batteries that store many GWH/TWH of power? Sure. Factories are already producing batteries by the TWH per year. That number is going to go up pretty rapidly as more factories come online. There's a lot of potential energy stored in (mostly) fully charged batteries in cars, grid storage, etc. That number is only going to go up.

A simple question: how much demand for seasonal storage is there actually? In gwh please. It's a simple question but I've never heard a coherent answer to it. As soon as you put a number on it, it becomes an engineering challenge that you can put a price on. Most of the reports on this front are based on very flawed assumptions. That's how you end up with stupendously large numbers.

I believe its too soon to to say gas plants are at the end of the road especially given that natural gas is often a by product of oil drilling.

A spanner in the works is Biden's tariffs that will hit grid storage imports from China, which is currently the large majority of deployed storage. These particular tariffs will start in 2026.

What about all the people saying that the longevity of these batteries is too short?

Too many new chemistries these days for that to hold true universally. Especially the sodium stuff.

Any such argument needs to accurately account for the relative ease of recycling lithium ion batteries, and the existence of batteries with significantly longer lifespans like iron air, compressed air (eg Hydroster) and pumped hydro

> relative ease of recycling lithium ion batteries

It's so easy there are no actual working recycling facilities for lithium ion facilities. Multiple projects to do so, including two only in Sweden, but nothing in operation.

So maybe it's not economically viable?

There are functioning recycling facilities, the primary problem is the lack of input batteries to be recycled. Right now the input is mostly defective batteries from production lines.

See for example Redwood Materials:

https://www.bloomberg.com/news/features/2024-04-18/redwood-m...

Given the expense of shipping materials long distances, it's quite likely that recycling will be far cheaper than processing from freshly obtained materials. But it remains to be seen what innovations will happen in the mining space or in the recycling space to change that balance.

What is the cost of recycling lithium ion, how long it takes? How much "iron air, compressed air (eg Hydroster) and pumped hydro" does cost?

It's not like lithium-ion batteries have a fixed number of cycles. It depends heavily on how you use them. At reasonable temperatures and keeping within 30-70% SOC they last a lot, lot longer than if you use them fully or at elevated temperatures.

For something like a big battery station, you can easily ensure the optimal battery conditions for long life.

And then there's using alternate chemistries as mentioned.

Such people are simply wrong/misinformed or spreading misinformation. The latter category is a significant factor as there is quite a bit of oil and gas money flowing into lobbying and misinformation campaigns. So just want to call that out as a thing. There are a lot of disingenuous voices in debates about this stuff pushing this or that agenda with fabricated horse shit.

But to summarize: this is not an issue. Not even a little bit. There are no nasty surprises involving batteries suddenly not working because they flaked out after a few months/years/whatever in the grid storage industry. This simply doesn't happen. Sure, individual battery cells might fail and then they get replaced or disabled. But there is zero indication that that is in anyway a frequent thing or a big deal in the industry. Mostly all we have is success stories. New batteries get deployed. News flash! They work exactly as advertised and operators save some money.

This is of course not surprising. Batteries used for grid storage are built to certain specifications. And that includes specifying how long they are expected to last (measured in charge cycles typically). So, all that is factored into the cost of the setup. Mostly all indications are that grid batteries are 1) very reliable and predictable 2) work exactly as advertised 3) might reasonably be expected to continue working quite long beyond their rated life time.

Besides, it is not like alternative methods are maintenance-free. I'd also hypothesize that battery installations are less complex than, say, a gas power plant, making the maintenance easier even if there is more of it.

> That’s helping to reduce system costs.

Costs go down, prices go up every 6 months. Can't explain that.

Please think about the poor shareholders..

Hey! There are pension funds in there, how dare you.

/end humour

That's exactly why they are there BTW, making sure that people's pension depends on the results of the financial market is an efficient way to prevent any action targeting the financial apparatus.

Now look up how much houses cost to build anywhere but the West Coast and maybe parts of the Northeast, and the cost per mile of California’s new high speed rail vs almost anywhere else.

It’s not a technical problem. The technology is fine.

Their example is California, where electricity prices are about 0.40-0.65/kWh.

Sole source advanced cycle natural gas plants are about 0.08/kWh.

That's quite a lot of cost.

I'm on LADWP and my rate is .19 per kWh, so you're off by 2-3x. Many municipal power companies are even lower.

If you're comparing using PG&E rates, I don't think a utility that has to pay off damages from causing one of the largest wildfires in history is a good metric for comparing costs of different sources of electricity.

I don't think the utility is the one paying off damages here.

They're paying significant damages. And it drives them to bankruptcy repeatedly. But at the end of the day there's just one grid.

Yes and then they turn around and immediately charge us, so it doesn’t seem like it is doing much.

There's no "turning around". You're in a place with a problem. The utility is not capable of wishing that away.

The utility could have fixed it with the money they got to fix it, of course.

PG&E has over 100k miles of electric distribution lines (excluding transmission). Their numbers state it's $3M per mile to bury a line. Did they get $300 billion?

60%+ of cost on my bill in northern CA is delivery, not generation. Which is dumb, cause it's a fixed cost charged as variable, but that's a separate discussion.

It's done like that to try and approximately apportion the cost based on each user's use of the infrastructure.

The cost is almost entirely driven by peak use though, so what would actually be "fairer" (or at least more cost-reflective) is to charge the transmission cost based on your individual contribution to peak demand on the infrastructure. People kind of hate that though because it's difficult for most consumers to predict or control their contribution to peak demand, and for the same reason it doesn't drive behaviour change / more efficient use of the infrastructure as much as you'd like.

In the end it'd probably still be about 60% of your bill anyway.

Some industrial users are charged in this way, and that's another driver for battery adoption as you can avoid the peak spikes with batteries locally in the same way the larger grid does.

FWIW where I live in Sweden the delivery is charged based on peak usage also for residential customers. Based on average of top three peak kWh/h in a month.

Iowa has more renewables as a proportion than CA yet has lower electricity prices than the national average. There are a few examples like this. Raw retail prices are not an accurate reflection of anything. You actually have to do your homework properly.

Iowa and California are on the same grid.

They're both the exact same resource mix because it's the exact same grid.

Here's the resource mix https://feature.wecc.org/soti/topic-sections/resources/index...

The difference in cost is entirely contracting and last mile transmission.

A quick glance at this next map will show that Iowa is not in WECC (although Idaho is):

https://www.wecc.org/_catalogs/masterpage/wecc/Images/WI-Map...

Also, just because 2 states are interconnected doesn't mean that the interconnects have enough capacity for electricity sources to be fungible.

You're right they're on the eastern interconnect, which is even larger and harder to reason about.

Edit: After the 2003 blackouts the Eastern interconnect got much more... connected

https://upload.wikimedia.org/wikipedia/commons/d/d4/UnitedSt...

Exactly. The price of Coke is based on where you're buying it not on the production cost.

The 0.08/kWh includes delivery, it's more like 0.04/kWh without delivery.

Well, that's just not what you are actually charged for delivery in California.

This article did not touch on it that well I think. Its less about replacing gas plants and more about replacing the peaking plants. I know this is true in Texas at least. A lot of sites are adding battery systems to replace an older peaker. Its very cost effective.

Seems more like .26-.46/kWh[0] And at a quick glance that seems more likely (to me) to be a response to capital cost spikes after all the wildfires as well as higher labor costs for the state, rather than because they are shifting to batteries[1]. The delivery layer is definitely bloated though, I agree with you there. Personally I think that's a regulatory capture problem.

[0]https://www.bls.gov/regions/midwest/data/averageenergyprices...

[1]https://www.bls.gov/regions/midwest/data/averageenergyprices...

I honestly think PG&E is going to collapse. It’s $0.49 / kWh here, which means our power bill would be $600-1000 per month if we didn’t have solar.

Full off the grid with propane backup costs less than $80K. ($20K each for panels, batteries, install and generator — I’m wildly overestimating intentionally).

At $600 per month, it takes 11 years for cutting the cord to pay for itself. More realistic estimates lower that to under 5-6 years.

Having said that, there are ecological reasons to give them free electricity from our batteries and excess solar.

However, that doesn’t change the fact that it’s cheaper to just build your own power plant and not do business with them.

Ultimately, someone needs to pay for grid maintenance, and, given the current insane market distortions, there’s hardly any economic incentive for individuals to continue to do so.

2000 kWh per month is crazy. Presumably that includes an electric car? Otherwise I don't even

Probably the potential differences in climate and home sizes between America and Germany.

2000kWh is not that exceptional for a normal American sized home.

Cooling (and heating) a 2.5k+ sq ft home with 10ft+ ceiling and open foyer will get you there easily.

If open foyer means what I think it means, an open foyer is pretty crazy.

Lots (most?) of 2 story US homes built in the last few decades have 20ft+ tall foyers or living room vaunted ceilings. Extremely inefficient for heating and cooling, especially since almost all have only 1 zone air conditioning.

https://www.homestratosphere.com/wp-content/uploads/2019/02/...

https://www.pinterest.com/pin/two-story-foyer-design-ideas-p...

We have a small, >4mi/kWh EV, the heating / cooking is all electric, and the house is small and extremely energy efficient by US standards.

Our panels produce 2000-3000 kWh per month in the summer, which is when we actually hit 2MWh of usage.

In California I assumed open foyer means literally open to the elements.

I should have been clearer and written foyer with tall ceilings where air flows freely between the top and bottom floor of the house.

Peaker natural gas plants aren't that cheap. Batteries are competitive with peaker plants.

Edit: misread comment, please disregard!

I didn't say that, it's from batteries and solar and wind.

Pretty in Texas is 0.02/kWh right now at wholesale. https://www.ercot.com/content/cdr/contours/rtmLmp.html

Seems lower than CA right now: https://www.bls.gov/regions/midwest/data/averageenergyprices...

I believe historically that hasn't always been true, but I'd bet that after all the bad press Texas utilities are eager to keep their prices as low as possible. In about 20 years I'm guessing we'll find out if that's because they've been skimming off their capital costs or what

There are a couple of these facilities being planned near me right now. They're receiving immense pushback from people concerned about lithium fires.

This isn't a realistic concern is it? I mean, compared to the pollution and fire risks of a plant that literally burns fuel when in use?

I checked out a Facebook group for a neighboring town's fight against this, and I don't think logical reasoning works. They're too busy praising Florida's HB 1645.

Yeah it's totally stupid. We already have fire risk from transformers, oil storage, and especially gas storage:

https://www.bloomberg.com/news/articles/2022-07-12/freeport-...

That’s a new excuse for nimbys, nice.

It's like the memory hierarchy. Always good to have a new level.

Needs a CDN

Reply on news.ycombinator.com