Hacker News new | past | comments | ask | show | jobs | submit login
A Solar Power Plant designed to deliver 100% of the electricity after sunset (pv-magazine-usa.com)
79 points by Osiris30 59 days ago | hide | past | web | favorite | 39 comments

For those not reading the article, it's battery storage. Specifically "Samsung SDI batteries". This is a great use case to see how well grid-scale battery storage will perform at the point of power generation in comparison to things like pumped water or thermal.

How do those batteries fare in terms of capacity loss over time ? My li-ion rechargeable sanyo eneloops AAA batteries seem useless after a couple of years.

It depends on the depth of discharge and heat management.

Laptops, phones, etc have a poor lifespan because they are frequently discharged from 100% to almost 0%. It is better to discharge them from 100% to 50% and then recharge.

Batteries generate heat during charging and smart phones can get hot during heavy use. Without good enough cooling the lifespan of the battery drops.

It also helps to charge them more slowly, so in case of solar farms, you can increase the lifetime of a battery by using more of them and charging them more slowly. Discharge is slower as well, which also benefits lifetime.

Usually the best case scenario is 80 - 20 as far as I know.

I've heard the opposite! Any piece of information that you can share to back it up? Highly interested.

Sorry I have studied this in other language and I not even sure how these things called in English. There is one graph that I remember called Woehler curve and there is another one that shows how much is the life expectancy operating between different charging levels (which I cannot find the name for).



It is called SN curve:


It shows a bit different picture for Li-ION that I remembered.

The reading I did years ago stressed that you should definitely cycle batteries to as close as Zero as possible, but if you want to do long-term storage keep them at 40%. Seems like there is a lot of myth around how batteries work and proper practice. A unified resource on this would be great!

Depends on the battery chemistry. NiCd and NiMh batteries which were very popular 10+ years ago do have a memory effect: https://en.wikipedia.org/wiki/Memory_effect

LiPo batteries do not have a memory effect, and you should avoid deep discharges.

I think the linked study gives you some insight. The problem is that we need all the capacity we get using a mobile phone for example. You can't really have a 2x capacity battery just for the ideal life time unfortunately.

I'm not a huge fan of the format of the article, but I am a huge fan of a solar plant that has sufficient storage (and mindset) to provide power when the sun isn't shining. The more like this, the better, imo.

That also irked me. I like memes probably more than the next guy, but the way it's been used here seems distasteful.

It might be more robust, cheaper and lighter on resources to just build a global electricity network, so there's always daylight powering the grid somewhere. Although for small, remote islands, batteries are probably a reasonable alternative.

The Pacific Ocean is pretty vast. When it's noon in Hawaii, there's not a lot of land area condusive to solar generation.

A global grid would be an immense coordination problem with tremendous costs. Some form of storage local to each grid is likely to be more cost effective. There are so many storage options, something should be appropriate as the need becomes real: pumped hydro / other mechanical, battery, synthgas (maybe), heat, pressure.

We have gone at most 580km/360mi with an undersea cable 95.8% efficient, €600m, 700 MW capacity https://en.m.wikipedia.org/wiki/NorNed

Repeat that cable 9.6 times over 3,459 miles from London to New York would be 66% efficient and cost €5.8b.

If energy is $0.05/kwh in London and you can sell it around the retail rate of $0.12/kwh in New York, your cable will make around $20k/hour ($0.028/kwh,after cable losses). Your cable will pay for itself after 33 years of complete capacity saturation.

With free energy at one side and retail prices on the other, the payoff time narrows to less than 8 years.

This calculation leaves out so much but it was a fun thought experiment.

Yeah, if we ever come to putting little artificial solar gen islands in the Pacific, nice clean synthetic gasoline being shipped from them would be awesome. (Generated from atmospheric carbon dioxide and water.)

Even if it only reduces storage demands it seems potentially worthy.

May I add a proposal to convert the mediteranian sea into a pumped hydro storage with long-term capabilities? It would need the same dam as the Atlantropa, but possibly less extreme.

There are some basins around northern Africa that might be amenable to slight terraforming, or just a re-activation of Operation Plowshare to blow a cavern which has it's rubble content lifted out and then used for subsurface pumped hydro.

Alternatively plan to collapse the cavern roof into itself after filtering persistent radionuclides and waiting for short half-life isotopes to decay. Collapse via conventional quarry techniques (a circle of drilled wells filled with ANFO?) or iteratively with robots that go out towards the center and drill to blast this off somewhat continuously. Then use a nearby water reservoir to pump in-between. Salt content shouldn't matter much except corrosion protection techniques.

Note that the cost per MWh is less than most fossil generation, and far less than the cheapest nuclear (per Lazard’s most recent LCOE analysis).

It’s clear that solar and battery storage can now provide affordable base load power, and can be deployed rapidly (this project is set to turn up in 2022).


LCOE is a poor measure of whether something can deliver baseload power cost effectively because it assumes all MWh are created equal.

What would make it clear whether solar + battery at this proportion is a viable source of baseload power is output volatility figures for a few years of operation (though one year would probably give a reasonable indication).

Tangent: The industry has a metric, availability factor, to describe what proportion of the time a plant can supply energy - but conventionally it only considers maintenance-related downtime and ignores periods where no energy is produced because of lack of 'fuel', giving figures of 99 and 100% availability for wind and solar respectively. That needs to change if we're going to measure whether renewables can provide suitable baseload power. The alternative, capacity factor, doesn't ignore periods of no generation in the same way, but it penalises dispatchable sources for having excess capacity so it's not great either.

Most power plants we use today can't deliver baseload cost effectively if you add in the externalized costs of climate change.

True, and I'd like to see this corrected with carbon pricing ASAP. My point is simply that whether or not intermittent renewables can ever cost-effectively deliver baseload power is still an open question.

Why do you think those costs are less than the existing taxes, or greater than what anyone could pay? Talking about externalities means you have a rough idea of a number, which is not "infinity". What do you think it is?

>Most power plants we use today can't deliver baseload cost effectively if you add in the externalized costs of climate change.

Like it or not tacking on the cost of climate change to power generation is so far from current reality that adding that hypothetical constraint is meaningless. The people making these decisions have real tangible costs that matter right now that they will prioritize in their comparison.

I'm not so sure about this. I wouldn't be surprised to see European countries introducing carbon taxes soon.

A lot less base load would be necessary if the electricity prices varied minute by minute based on the cost to provide it. This is because a great deal of electricity use can be easily time-shifted, but people don't bother doing that because the consumer price is the same 24/7.

What would be even better would be if power-producers and distributors (or, more likely, aggregating intermediaries) made available some APIs that provide real-time data on pricing, sources of power and a degree of "peek a little way into the future" prediction. Then we could build/install devices that take advantage of that data. e.g. A fridge could heuristically trade off power price, next-hour likely pricing, source (maybe I prefer green energy or energy generated 'close by', and am willing to pay some premium; maybe I don't care and just want cheap), internal temperature and local time-of-day usage patterns affecting temperature management. (I chose 'fridge' as the example because its energy consumption can be time-shifted, but only to some degree. Eventually it is going to want to switch on regardless of energy supply conditions.)

On the flipside, appliances (water heating, fridges, cold-rooms, space-heaters,e tc.) should be encouraged to talk back to the suppliers APIs to indicate likely near-future demand, current constraints and demands and so on, so informing the 'grid' with demand data at micro-scale.

I guess this is something like what people might mean when they speak of making the grid 'more intelligent'...

Ironically this is one problem which is incredibly well suited to a market based solution, but the lobby groups which typically advocated for "let the market decide" (e.g. heritage/CATO/Americans for prosperity) took a lot of their money from fossil fuel companies. While "let the market decide" was a workable message when fossil fuels were cheaper than renewables, it no longer works for them so they no longer advocate for it, so it's not an idea that is pushed very hard.


It's incredible how the Australian government goes on endlessly about saving money but continues to ignore the fact that renewables are cheaper.

Guess that's the power of political donations

I guess it's not the public money they're trying to save!

Here's the vendor datasheet for the batteries: https://www.samsungsdi.com/upload/ess_brochure/201902_Samsun...

It doesn't say the specific chemistry but it shows a chart estimating 6000 100% charge/discharge cycles at 1C until it hits 80% capacity.

So for this system that'd be 16 years of full cycles till it wore down to being "only" a 240MWh system.

In reality they'll probably cycle it less aggressively than that (95/5 or 90/10) to meet the expected 25 year lifetime of the overall system.

I'd just like to point out that there are solar plants that can generate power after sunset too. Solar thermal plants can store their heated working fluid in an insulated container and use that to power their heat engines later on.


CSP is no longer cost-competitive with PV for raw energy cost because it relies on the same steam turbines that coal and nuclear stations use. PV is now cheaper than those turbines in much of the world—in some cases, so much so that even with batteries, it can beat CSP.

I thought the article was going to be about these until I clicked on it.

Layman Science Wizard recently explained to me that pumping water up into a storage tower (using solar-based energy) and then using the gravity-drop to generate electricity again (when the sun is gone) was also very effective. A natural battery of sorts. No batteries to wear out, neither. Feasible?

Just to point out a subtle distinction I hadn't caught reading the headline: the station is designed to only deliver its power at night. That's the 100% in the headline. The battery storage apparently lasts 7 hours, so there's still a good five hours of the average night not served by the plant.

Great. Let's build a million of them.

> Engie says that the Samsung SDI batteries which it has chosen for the energy storage solution will be able to deliver electricity for up to seven hours

For anyone who whats to know what SDI stands for it is "Samsung Display Interface".



Are you a fan of copy pasting posts?

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact