Hacker News Reader

Comment by manfredo | Hacker News Reader

A few things stick out to me:

* This study assumes that we can ramp up purchases of solar and wind generation to match this graph: https://imgur.com/lMU252r while costs continue to decrease. That's not how markets work. Rapidly buy solar panels and the cost of solar panels go up. Put solar panels in all the places where it's easy to install and now you have to pay extra to put them in places where it's not so convenient.

* This study thinks we can achieve this with only ~200 MWh of storage. This is an optimistic projection. To put this in perspective, daily US energy use is 11.5TWh. We're giving solar and wind generation only ~15% allowed variability before we're dipping into gas. Yeah, it does say that it's only 90% clean electricity, but even reaching this figure is optimistic.

View on HN

That's far too optimistic about storage. For small off-grid systems, you need ~5 days of battery capacity to ride out lack of wind/sun. On a large scale you'll need less because you can move power from place to place.

But still. Sun is down half the day and wind speeds can be low in areas hundreds of miles wide. Its obvious they will need at least a day of battery capacity, probably more. Not hours.

Solar and wind are great. Many areas in the US have hours long periods during the day where power is 100% renewable. But batteries are way too expensive and I don't like when politicians make laws based on predictions 15+ years in the future.

How did you get 200 MWh? From page 20 of the report:

"In summary, retaining existing hydropower capacity and nuclear power capacity (after accounting for planned retirements) and about half of existing fossil fuel capacity, combined with 150 GW of new 4-hour battery storage, is sufficient to meet U.S. electricity demand with a 90% clean grid in 2035, even during periods of low renewable energy generation and/or high demand. Under the 90% Clean case, all existing coal plants are retired by 2035, and no new fossil fuel plants are built beyond those already under construction. During normal periods of generation and demand, wind, solar, and batteries provide 70% of total annual generation, while hydropower and nuclear provide 20%. During periods of high demand and/or low renewable generation, existing natural gas plants (primarily combined-cycle plants) cost-effectively compensate for remaining mismatches between demand and renewables-plus-battery generation—accounting for about 10% of total annual electricity generation, which is about 70% lower than their generation in 2019."

150 GW of batteries with 4-hour discharge is 600,000 MWh.

http://www.2035report.com/wp-content/uploads/2020/06/2035-Re...

The graph displays 150 GW, with no duration specified. 4 hours of 150GW storage, or 600GWh of storage, is still only 5% of the nation's current daily electricity consumption.

And that's probably going to be an even smaller share in 2035 as more vehicles go electric.

> This study thinks we can achieve this with only ~200 MWh of storage.

Agree, very unrealistic.

One thing that always disappointing to me about these conversations is that solar == photovoltaics, batteries == li-ion electrochemical… luckily the world isn't limited to these asumptions and developments have been and continue to be made, even if they are mostly ignored/overlooked by the HN crowd.

Things like kinetic storage are geographically limited. Most places with available hydroelectric storage already generate electricity from dams.

Chemical storage in the form of natural gas produced by the Sabatier process is very inefficient. Currently it's at 55% efficiency to go from power to gas. Even the most efficient combined cycle plants are currently 65% efficient. End to end efficiency is ~1/3.

Also, using synthetic natural gas requires a means to capture the exhaust of gas plants to actually make it carbon neutral. Converting the fraction of a percent of carbon dioxide in the atmosphere is not currently possible.

The conversation is mostly about batteries because that is the only known option at the moment, other than geographically limited ones.

There's also many kinds of thermal energy storage systems in use (I think molten salt systems are the most promising[0], esp if you design salt mixtures that have have higher delta k's[1] and more efficient HEX and steam turbines to about ~60% efficiency of converting heat to electricity[2]), not limited by geography.

Going of the example powering 100MW turbine with a working temp range of ~278C[0] and using an eutectic mixture of calcium chloride and magnesium chloride (ionic halide salt)[1] with a working temp range of ~1440C, can make the "tank of about 9.1 metres (30 ft) tall and 24 metres (79 ft) in diameter to drive it for four hours by this design." last for nearly a day, though with higher upfront cost than a typical NaNO3 + KNO3 system, but nowhere near nuclear costs (except for the similarities with dealing with supercritical water for electricity generation).

[0]: https://en.wikipedia.org/wiki/Thermal_energy_storage#Molten-...

[1]: https://gist.github.com/cinquemb/015f8b0775db2381513f0e20186...

[2]: https://www.solarpaces.org/vc-funded-thermal-battery-based-m...

Solar thermal plants [1] effectively already use this system. They heat a model salt container, which drives a heat engine. The molten salt remains hot enough enough to keep producing energy through the night. However this scheme requires both solar collectors (in the form of mirrors) as well as the molten salt tank plus a heat engine. The resulting levelized cost is about the same as nuclear power [2].

If this startup does manage to produce energy storage at the efficiencies and costs as advertised that'd make intermittent sources more viable. Ultimately building solar and wind is betting that we'll be able to build efficient, cheap, and massively scalable energy storage.

1. https://en.wikipedia.org/wiki/Concentrated_solar_power

2. https://www.lazard.com/media/450784/lazards-levelized-cost-o...

Yes, I had CSP in mind, however most CSP systems today use the kn03+ nano3 eutectic mixtures, and the LCOE is based on that assumption, as well as assuming you can have no baseload from the storage, and cannot distribute storage separately from collection. And even with their assumptions the range for solar thermal capital costs/kwh is about 20% less than nuclear on the top end, 41% less on the bottom end. One also has to consider the political costs that nuclear comes with and that CSP doesn't.

I seriously doubt America will buy solar panels quickly enough to outstrip supply. Much more likely that production will increase and costs will go down due to economies of scale.

Also there is no shortage of places to put solar panels, especially in America.

Are you predicting that solar panel production won't continue to scale to meet demand?

The limiting factor of producing solar panel at these immense scales is the cost of raw materials. As other posters pointed out this study is already assuming that the cost of solar panels and wind will be barely greater than the cost of materials.

And yes, if the US tries to increase it's solar capacity by a factor of 50 in the next 15 years (and especially while other countries try to do the same) these materials will become more scarce and thus more expensive.

Reply on news.ycombinator.com