Hacker News new | past | comments | ask | show | jobs | submit login

Battery storage is not an option for large scale. Battery technology is good for bridging gaps measured in minutes while larger power plants spin up and react to demand. That's a real benefit, and helps to eliminate some expensive peaker power plants, but the technology is simply not even close to being able to fill in the gaps needed for a grid where large percentages of the energy is wind or solar.

Nuclear is currently the only technology that exists that can do this.

The back of my envelope disagrees. I'm distracted so LMK if there's a major flaw in this analysis.

Nuclear power generation costs ~$6k per kW(1)

Solar power generation costs ~$1k per kW(2)

Solar capacity factor is ~25%, so ~4k/kW to compare with 24 hour baseload power.

Storing 3 kW for ~12hrs requires ~36kW-h storage.

Li-ion battery packs are getting to ~$100/kW-h (3) 36 kW-h storage is ~$3.6k

Nuclear cost: ~6k per kW baseload

Solar + battery cost: ~7.6k per kW baseload (~4k/kW generation, ~3.6k/kW storage)

(1) https://www.world-nuclear.org/information-library/economic-a... (2) https://www.nrel.gov/docs/fy19osti/72133.pdf (3) https://cleantechnica.com/2018/06/09/100-kwh-tesla-battery-c...

Lithium deposits are already facing depletion and the price has been steadily rising for a decade now with the new demand for li-ion batteries.

Building enough battery capacity for the grid would be far more than current world demand for Li-ion cells. I imagine prices would skyrocket, throwing off your calculations.

Grid level battery storage is very likely moving to other chemistries. The only reason to use lithium is that it's the chemistry that's the most developed right now, but lithium-ion batteries do not fundamentally speaking have the best properties for grid storage. Lithium-ion is great for energy density, but that's not a critical requirement for the grid.

For grid storage I think molten metal makes the most sense. It's a technology that was developed to be ideal for grid storage from the start. Flow cell batteries might also make sense. And then there's other storage technologies like compressed air, pumped hydro, storing kinetic energy, storing thermal energy, etc.

http://news.mit.edu/2016/battery-molten-metals-0112 https://www.youtube.com/watch?v=NiRrvxjrJ1U

Lithium prices have gone down almost 60% in the last year.

Short-term storage isn't that bad, we have other battery chemistries or even completely different types (e.g. flow batteries). A bigger problem is seasonal storage, for which most batteries are far too expensive.

The fears of peak lithium are greatly overblown. New mines are coming online now that the element is in higher demand and the price is going down.

But that's only one of many energy storage technologies that can be used for batteries. It's popular for mobile/portable uses because it has high energy density, but that's not really necessary for grid-scale energy storage; whatever's cheapest will do (which may well be pumping water uphill).

Lithium deposits?

Most of the lithium found today is extracted from brine reservoirs located in regions of southwestern South America and China.

> Nuclear cost: ~6k per kW baseload

> Solar + battery cost: ~7.6k per kW baseload (~4k/kW generation, ~3.6k/kW storage)

So nuclear is cheaper than solar + storage?

China, South Korea are $2K-2.5K per kilowatt of baseload. A Nuclear kilowatt generates 70-95% of the time. A solar kilowatt generates 10-20% of the time. A Gigawatt of solar generates 1 Terawatt hour per year. A gigawatt of nuclear generates 6 to 8 terawatt hours per year. So nuclear generates an average of 7 times more. Solar lasts 15 years. Nuclear lasts 40-80 years. Solar needs to be rebuilt 3 to 5 times versus nuclear. There is no supply chain for matching battery storage at scale. They are just starting to build batteries for cars at the 200 gigawatt-hour levels. Solar and wind in California and many other places has only 10% of the generation during winter. There is no 90-day power storage and building one would be insanely expensive.

You pay for electricity by the kilowatt-hour.

Don't forget LiIon batteries only get a few hundred recharge cycles. Is that calculated in this kw-h figure?

Utility-scale power storage can use other technologies, such as pumped-storage hydroelectric generation. Basically they pump water uphill as a means of storing energy.


Pumped storage is not realistic. It’s a cute nice to have on the side but cannot ever possibly scale enough to meet even double digit percentage of our need. To be quick about it let’s take some data from [1].

Hydroelectric power generates 6.1% of all US power today. All those huge dams you see everywhere with their giant lakes you can see from space that did massive destruction to ecosystems across this country? Those generate a measly 6.1%. (Blows my mind I didn’t even know it was that low.)

Even if you turned every hydro dam in the US today into a pumped storage facility it would be barely a curiosity on our energy needs.

And you sure as heck aren’t going to 10-fold increase the number and size of dams and lakes we have in this country. Nobody will stand for that.

We all seem to keep doing these wishful mental gymnastics to try avoid nuclear power, but the numbers just never add up.

1. https://en.m.wikipedia.org/wiki/Hydroelectric_power_in_the_U...

What you're not taking into account is that hydroelectric power production facilities are currently sized only for the gravity-driven amount of water that reaches their reservoirs. Without making the reservoirs any larger, or adding any additional dams, you can simply add lots more turbines and pumps to the existing reservoirs and get a lot more power production out of them.

Consider that the vast majority of the pumping would be to even out the daily power cycle (the "duck curve"), whereas reservoirs are sized to hold years' worth of water. The amount of water pumped back uphill during the peak solar output of the day would be a negligible amount of water to the overall reservoir, and then you'd run it down through additional turbines in the evening to produce power.

Following up on this, here's some math.

As you may know, Lake Mead (the reservoir for the Hoover Dam) is currently running very low owing to various water shortage issues. If you've flown into Las Vegas recently this is very obvious. It's currently at only about 40% of its capacity, which is a shortage of about 210^13 L. The Hoover Dam's hydraulic head is 180m at peak height, but let's call it an average of 160m for our purposes below. Using the equations here: https://www.engineeringtoolbox.com/hydropower-d_1359.html

For the total amount of energy available if we were to use solar to pump the reservoir up to full during each day and then generate power at night:

PE = (1 kg/L) (210^13 L) (9.81 m/s^2) * (160m) = 3.14 * 10^16 J = 8.72 * 10^12 watt-hours (this should be knocked down a little bit for efficiency losses; cursory Googling shows that turbines are roughly 90% efficient at turning PE into electricity). Contrast this figure with the annual total electrical usage of the entire US of 4 * 10^15 watt-hours. Divide by 365 and you get 1.1 * 10^13 watt-hours.

So, if you fully pumped just Lake Mead up to its full capacity and then ran it back down its current level each day, you could store most of the energy used by the entire country in a day. Just in that one reservoir. Obviously you'd need to add a lot more pumps and turbines to do so, like orders of magnitude more, but the point is that you wouldn't actually need any additional land to do so. If you're willing to fill up and then empty Lake Mead each day, you can easily do more than the power requirement of the entire country.

So anyway, that's a long way of saying, yes, pumped storage is entirely realistic. Add in all the additional extra capacity in other existing reservoirs across the US and you can easily store many days' worth of power in reserve, just using pumped water.

That's an awesome thought :) but as you say, now we need to also get a few hundred GW of generating capacity out of the Hoover Dam and then we'd have something. Hoover Dam is about 2GW nameplate capacity IIRC, so if we could now somehow dig out 199 equivalently sized new turbine halls underground around the dam we would have a real tourist attraction. That might look something like 398 Manapuori power stations. We would also need to install however much solar is required to both provide enough renewable power during the day to offset fossil sources, and have excess to pump enough water to store energy in our Extreme Hoover Dam project to power the country during the evening post-solar peak hours and through the night.

Is there a good study that explains how pumped hydro and solar can actually work to make a significant dent in our gas/coal power?

Some more spitballing:

Demand ranges between 400-650GW over a summer day (over 700GW in heat wave). If we look at EIA data for a summer week we see Hydro produces about ~50 GWh at peak, ~21GW at a low point, over a day. And we see fossil sources producing about ~270GW at minimum to ~460GW maximum over a day. Solar producing nothing at night up to 22GW then unfortunately falling away too early to contribute during the peak demand period (see the duck curve).

So the argument for pumped storage here seems to be that we can somehow get that 21GW to 50GWh production up to some meaningful number. Lets assume we can convert every dam in the country into pumped storage (obviously not but let's assume). Now as discussed need to increase the production capacity of hydro a lot. Let's say we can quadruple the generating capacity of every hydro dam in the country and turn them all into pumped storage. 200GW would be meaningful (not a full solution but nearly half way to a solution).

How? Sounds incredibly unlikely to me. Especially given not all dams are well suited to pumped storage anyway. You build new tunnels and pumping systems to get the water from downstream lakes back up. You add three more generating halls for every one, probably buried alongside the dam, how much is that going to cost? A lot. How long is it going to take? A lot longer. We need something that we can production line produce at this point.

Now if we wave a wand and somehow do that though, we could produce a maximum 200GW with our hypothetical hydro/storage set up. But we now also need to build however much solar is also necessary to reach our green 200GW target and pump that water back up during the peak solar period so the hydro can run through the non sunny part of the day giving us some 200GW of continuous Solar+Pumped Hydro base generation. That would have to be somewhere in the vicinity of what? I'm spitballing but maybe like 400GW of solar we need to install? How much do we add on for that cost? So we've quadrupled our dam's generating capacity at some incredible expense and built on top about 13x the amount of solar we currently have installed.

And we still have to keep fossil around to generate 70GW at night and 260GW during the day.

I just don't see how we get pumped hydro beyond anything more than a curiosity at this point. (That doesn't mean I think it shouldn't be pursued where it's feasible and the business case stacks up.)

You seem to know what you are talking about based on your comment history, however this statement needs some backup to be accepted at face value:

> All those huge dams you see everywhere with their giant lakes you can see from space that did massive destruction to ecosystems across this country?

"dams are big" x6

You also need to consider that solar/wind+battery needs to be massively overbuilt to get reliable power because of the variability, both day to day and seasonal. More sophisticated analysis needs to recognize that as your percentage of power from solar/wind approaches 100%, the amount of overbuilding required rises very rapidly.

See e.g. https://www.technologyreview.com/s/611683/the-25-trillion-re...

Only if you do it as proposed in the battery-centered article.

An important point is that renewable energy and battery storage continue getting significantly cheaper with every passing year, whereas nuclear power is not getting any cheaper. That's why I said we're at the inflection point. It's not accurate to just take today's figures and project them forward; anticipated gains need to be factored in too. That's why nobody's building new nuclear; even if we assume your numbers are correct and say it's slightly cheaper now, it won't be in a few years. Nuclear power plants need to run for decades to recoup their large capital construction costs.

Tesla Powerwall: $14.5K for 27 kwh and $2.5-4.5k installation. $6.8K for each 13.5 kwh of Powerwall.

Batteries by themselves not enough. Need inverters etc... You cannot get to the cost of the car by only adding up the cost of gasoline used over its lifetime.

China's and south Korea and Russia have nuclear build costs in the $2k-2.5K per KW range. They make 70% of the world's nuclear reactors.

Nobody was talking about Tesla Powerwall. We're talking grid scale energy production and storage, not at the level of individual houses. You can't remotely compare a nuclear reactor to a house-level battery setup.

What is the lifecycle on the batteries? What about the externalities of heavy mining to find the materials for batteries? Isn't it really toxic? Is it even feasible to run the whole world immediately on rechargeable batteries? How much more environmental impact would be levied, as opposed to nuclear plants?

Everyone says that, but no one who says that shows us their math or cites any sources. Show your work. Don't just say "can't". Consider variations in both output and load, both of which are well documented. Consider cost, which is also documented. Consider change in cost, which has been considerable.

Don't just say "can't" without any justification except your sense of certainty.

Power-to-gas technology can possibly fill that niche.

Power-to-gas would be a cool technology if we could figure it out, though honestly if we did it would make the argument for nuclear even stronger. Whatever the technology ends up being, it will benefit from large, efficient facilities fed by large amounts of clean, steady power. That means nuclear if you want the overall cycle to be carbon neutral.

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