Hacker News new | past | comments | ask | show | jobs | submit login
Crescent Dunes, a utility-scale solar power plant in the Nevada desert (time.com)
44 points by prostoalex on June 19, 2016 | hide | past | web | favorite | 37 comments

"All night" isn't quite true. It has 10 hours of storage. And if you want full capacity, you have to start tapping into that in the evening.

Also what's cool about this plant, is that it doesn't use natural gas in the morning to heat up the molten salt (like other plants). The salt never cools completely during the lifetime of the plant.

Expensive though.. on a 25 year contract at 13.5 cents per kwh. That'll be renegotiated after the 25 years at a much lower rate. Utilities pay higher prices for new plants (gas, oil, etc) to pay off the cost of building it. After that, it drops to a more reasonable price.


Doesn't "hours of storage" necessarily imply a rate of discharge? If you don't take any steam off it, doesn't it last more than 10 hours? It seems like they should just state the storage in terms of extractable energy stored.

And why sodium? It does not have a spectacular specific heat. Is it just convenient due to its high boiling point?

Yes.. they do state the storage in terms of energy. When plants state hours, it's at full capacity. So 110MW capacity x 10 = 1.1 GWH of storage. Of course it will operate longer if you use less than full capacity.

Molten salt is used because it's liquid at atmospheric pressure, is low-cost, its operating temperatures are compatible with the most efficient steam turbines, and it is non-flammable and nontoxic.

I believe it's because of the large temperature range between solid and gaseous phases, there is a ~800 degrees Kelvin difference between the boiling and melting points, which at a specific molar heat of 29 J/mol/K is greater than the energy to heat water to boiling for a ~100 degree difference at ~56 J/mol/K

Expensive indeed. I wonder how concentrated solar thermal compares with a simple photovoltaic plant these days? Given the precipitous drop in price for PV panels, even if solar thermal is currently cheaper, how much longer until it's just not competitive?

PV is down to around 5 cents/kwh. I seem to remember a new plant recently that was 2-3 cents, but can't find it.

The only problem with PV is its variability. When there's cloud cover, in the evening, etc.. power drops. Which means you need some kind of storage to even it out and probably a gas plant for the evening/night time.

In comparison, CSP has a steady output, which lets them make adjustments easier.

Much of the variability problem could be eliminated by having real-time pricing of the power. Lower prices when the sun is shining, higher prices when cloudy. Then, build into usage devices a monitor for the price, and those that can shift usage (such as charging the car and running the electric water heater) can not suck power when prices are higher and draw power when it is lower.

I.e. solve the problem on the demand side rather than the supply side.

I think this is going to be a massive part of the new energy economy. I'm picturing the power grid becoming less of an 'energy retailer' and more of an 'energy broker', with each property's power control module placing bids on energy when it needs to import and offering sale of energy when it has a surplus. There will then be a flat rate (or close to, probably tiered based on peak consumption) for access to the grid, and the power utility company becomes a power grid maintenance company rather than a generation company.

Sounds like a hybrid plant would be the best approach, with a large molten salt reservoir which is 'charged' while the sun is shining and used to run a steam generator when PV isn't available. Of course, the way battery storage prices are going, I wouldn't be surprised if battery backup becomes more economical than all that messy pipework in the near future anyway.

I don't share your optimism. It'll be a while before we can produce useful, grid sized batteries for this kind of use.

The hybrid approach you describe would require twice the surface area and probably wouldn't double the output.

I think I prefer the idea of having large plants like these, with 'night'-time capabilities and a relatively steady output. That sounds like it would work pretty well in combination with decentralized roof top solar + smallish home batteries.

The whole point is that solar thermal power costs 2x - 5x what solar photovoltaic power costs, so you use solar photovoltaic when possible and back it up with solar thermal. It does increase the area required but that's the price you pay for fully renewable baseload power. You could just use thermal but it'd be much more expensive and still require the same footprint for similar power output.

Of course, there's nothing wrong with the photovoltaic part being distributed in the form of rooftop solar, which is what's already increaslingly happening. Home storage batteries for load leveling are picking up nicely too, which is good because they're going to become mandatory within the next decade if we want metropolitan power grids to survive tens of megawatts of sometimes-rapidly-fluctuating solar power being dumped into them.

> the way battery storage prices are going,

And we didn't even start playing with Sodium as a replacement for Lithium in those batteries.

Batteries have two things going for improvements in the batteries themselves and more energy efficient homes and appliances. PC won't replace utilities in densely inhabited regions, but batteries may help evening out variable sources.

A hybrid plant isn't possible. Apart from both producing electricity from sunlight, a PV plant has nothing in common with concentrated solar thermal.

"A hybrid car isn't possible. Apart from both producing torque, an electric motor has nothing in common with an internal combustion motor."

Hybrid just means that it uses both approaches. I'm suggesting a solar thermal plant which is generally lightly loaded or unloaded during the day, storing the heat in a large reservoir while a neighboring PV plant much more cheaply covers demand. Then when the PV plant can't generate due to cloud cover or night time, the thermal plant steps up and uses the stored heat to generate power.

It's no different in principle to a hybrid car where a battery electric drive system cheaply covers as much driving as it can, and an internal combustion engine kicks in when required.

There has been at least one combined PV + CSP scheme proposed [1][2] in Chile.

This article [3] analyzes some of the merits of this approach.

[1] (Partial paywall) http://social.csptoday.com/technology/integration-csp-and-pv...

[2] https://en.wikipedia.org/wiki/Copiap%C3%B3_Solar_Project

[3] http://spie.org/newsroom/6018-photovoltaics-and-concentratin...

> "All night" isn't quite true.

Ok, we've replaced that title with representative language from the article.

Its great these things are coming online, they have issues when the mirrors get out of alignment. From lighting the lower parts of the tower on fire [1] to blinding pilots flying nearby [2].

[1] http://www.latimes.com/local/lanow/la-me-ln-solar-plant-fire...

[2] http://www.geek.com/news/worlds-largest-solar-power-plant-is...

This is really two things in one - a solar plant, and an energy storage system. Couldn't we decouple them? I.e. when electricity is cheap (Germany during the day can dip to zero), heat up the salt, and discharge it at night (actually, in the evening, when electricity is most expensive).

Also, I wonder how it compares to big batteries?

This is not a photoelectric plant. It is a sunlight concentrator to heat salt which stores energy which is eventually used to drive a steam turbine and turn a generator. All one thing.

You could use molten salt to store electricity on its own, but I think the efficiency would make it undesirable. A steam turbine is something like 50% efficient and is probably the least efficient part of the system, so it just gets worse from there. For comparison, lead acid batteries are above 70% efficient.

But if electricity is free during the day (I guess because of oversupply from wind and/or solar?) who cares about a 50% loss?

I wonder if using the heat directly e.g. with a Stirling engine/generator would be any more efficient?

Lead acid batteries are an environmental nightmare compared to molten salt.

Electricity wouldn't be free during the day if you wasted 50% of it.

No doubt they learned a lot from Gemasolar. The article describes Crescent Dunes as the first "utility scale" plant. The wikipedia page is a bit more quantitative, listing its capacity at 125 MW, vs Gemasolar's 20MW:


I've started to become more and more convinced that the longer term view of the power grid is that it may no longer supply an assumed base-load. Instead every electronic device will end up having some measure of chargeable battery backup along with the building/facility the device is in. Over a day the batteries will soak up power and charge and at night they'll power the devices.

Current thinking about power grids is that power generation is cheaper at night because people are using it less. But in a solar powered system, power generation during the day is essentially free, it's the capture devices (and storage devices) that cost anything.

This notion helps the power companies share and distribute the storage costs while making the direct users of the power responsible for sizing their storage costs.

I'm sure they researched this but square panels seem suboptimal. I wonder if manufacturing made that determination. It seems like circular or hexagonal panels would better utilize the space. That or just grid aligned square panels. Maybe the space efficiency doesn't matter that much? Each configuration captures the entirety of indirect sunlight. The little bit extra you'd get from direct sunlight is likely insignificant compared to overall performance.

I don't think the panels are the limiting factor. Remember this isn't a photoelectric plant; the sunlight isn't being converted directly into electricity. All the panels are doing is reflecting sunlight to the central tower, where it heats up molten salt. The limiting factor for power generation is probably the heat capacity of the salt, not the amount of sunlight the panels can reflect.

Also, the panels don't look like flat squares to me; they look curved, so that they concentrate the sunlight they reflect.

To manufacture circles you'd probably be making squares then cutting off ~1/4 of the area and throwing it away.

I don't think space efficiency is much of a consideration. These things tend to be built out in the desert where all you've got is space.

Circular panels would be a terrible idea, circles can't tessellate.

http://www.wired.com/2016/05/huge-solar-plant-caught-fire-th... has a decent summary of the problems with this type of solar.

I was surprised that the Ivanpah fire wasn't mentioned in the Time article but then it seemed pretty rah-rah.

I drove by this plant to/from Further Future. The concentration of light around the central towers was so intense, it was visible - a light, glowing cloud (clearly in the "shape" of the reflected light) surrounding the central tower.

If transmitting power was more efficient we could just have solar in the lit time zones powering those in the dark.

Exactly. With our current techniques -- high-voltage or superconducting lines -- as I understand it, it's more expensive to transmit than to generate "locally" (within a few hundred miles?). I'd be curious to see the tradeoff (1) researched better and (2) visualized quantitatively.

Sorry I don't have exact numbers for you, but there do exist some very long HVDC lines [1]. The longest as of writing is in Brazil at 2385 km. There is a fairly long one in the US at 1360 km between the Columbia River Gorge and the Los Angeles basin [2].

But given that the solar terminator moves at about cos(latitude) * 1668 km / h (so slower towards the poles), even at a generous mid-Europe latitude of 45° N it'd be moving at 1180 km / h, giving you only a 2-hour window to make use of such a scheme with the HDVC lengths we have now. But hours before sunset, you're probably not at peak solar production anyway.

[1] http://www.power-technology.com/features/featurethe-worlds-l...

[2] http://new.abb.com/systems/hvdc/references/pacific-intertie

That's not the only problem. You'd have to get the cooperation of many different countries who don't necessarily like each other. The political obstacles would be greater than the technical.

I think there's one being built in Israel too?

Registration is open for Startup School 2019. Classes start July 22nd.

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