Comment by Robotbeat | Hacker News Reader

Comment by Robotbeat | Hacker News Reader

Using nameplate capacity versus actual delivered kWh is a little misleading. For instance, 1GW of nuclear can out-produce 3-4 GW of solar, and it does so at a constant rate, not correlated with all the other solar farms.

I'm hugely bullish on PV long-term, but it's important for us to keep this sort of thing in mind. And the discussion of the /peak/ output of solar or wind as a percentage of total electrical output is also not terribly helpful.

By the time solar accounts for 10-15% of total energy production on the grid, there will regularly be times when solar is providing almost all the grid's power. But overall, it's still just 10-15%.

In order for solar to displace most of the energy produced on the grid, it's necessary for solar to actually produce 2-3x the actual end-use electricity demand at times, with the output stored in batteries or sometimes just wasted.

And according to my calculations, in order for solar to provide for 100% of the power produced even during winter (without requiring weeks of storage), you're going to need a nameplate capacity approximately 20-30 times the average grid consumption (on an energy basis, oversizing the array by a factor of 3 to 5), and most of the energy will need to be thrown away. This is obviously going to be more expensive per kWh, but this is why a mix of power sources is essential for economic power production, and at high latitudes (like the UK and Germany), this is even more important.

(Note: it is possible to improve that oversizing factor if you have extensive HVDC power lines and hydrogen production, but those things are expensive and PV is cheap, so often it might still make sense to just over-size the array.)

> and most of the energy will need to be thrown away.

I won't speak to the veracity of your numbers. I have no idea your methodology or even if you're qualified to do an analysis here or the quality of your data, so...

But I will point out that the current plan to address this is to repurpose and recycle existing lithium ion battery packs at sclae, while continuing to develop very large electrical energy storage methods.

A real-life example of this is Tesla's Australian contract (https://www.reuters.com/article/us-australia-power-tesla/tes...) which uh... literally started on twitter as a dare between rich people (https://twitter.com/elonmusk/status/840032197637685249?lang=...).

Very large power storage stations like this actually offer a lot of utility in a variety of scenarios. Not only do they make distributed grids much more reliable offer a facility to ground out excess energy on the grid (helping reduce the impact of solar phenomenon over large-scale transport networks), but they also have a lot of value in disaster recovery scenarios.

> "...but they also have a lot of value in disaster recovery scenarios."

How so?

I'd imagine that you'd need a pretty large capacity battery storage system in order for it to reliably act as a blackstart generation source.

EDIT: I did some quick research, found this recent article about IID's blackstart battery testing - https://www.energy-storage.news/news/california-batterys-bla...

Thank you for linking this, I neglected to do so in my original post.

Sure, my calculations assumed about 48 hours of storage. But there's a pretty big difference between 48 hours and seasonal storage where you need weeks (400 hours).

What Tesla is currently doing is installing about 4 hours of battery for the typical site, so about two orders of magnitude away from being able to scavenge all that wasted power on a fully PV grid, but fairly optimal to ensure full use of stored power.

It is key to understand the seasonal variation of sunlight. This is a much worse problem at high latitudes. But at lower latitudes (say, anything less than 45-50 degrees, i.e. >80% of the world's population), PV is now cheap enough it's cheaper to just over-size the PV array to provide power even when it's cloudy and/or winter than to provide weeks or months of storage.

EDIT: Maybe "wasted" is the wrong word. Is your car engine "wasted" just because it's not red-lining 24/7? Averaged over a year, your engine only has a "capacity factor" of about 0.2%, but the energy is there when you need it, and it's still affordable. Likewise with solar, it's okay if not all the electricity is utilized, as long as it produces enough on cloudier days to not require a ton of storage (up to ~50 hours of storage is fine).

Forgive me, but it seems to me like you didn't read the materials that I passed you in instead based your commentary on Tesla's powerwall product?

Ive never understood why they use nameplate, it should be possible to predict t fairly accurately the yearly output based on latitude or whatever.

Though note that something similar works against solar PV sometimes as well e.g. people will say something like "we use X amount of energy a year, to replace this we need such and such an area of PV.

Often these calculations don't take into account the percentage of coal, gas or oil energy which is thrown away as heat in order to get generate electricity, or move vehicles. (Even when generating heat, electric heat pumps can often do better than 100% by moving it around instead).

That can make PV 3 or 5 times better than it seems. EROEI calculations in particular are bad for this.

It's a direct vs indirect measure, and can be directly compared between sites (regardless of capacity factor). It also more directly represents costs.

The general problem is common in metrics and assessments.

It takes a lot more than just the latitude, but we can make a very accurate prediction. I once worked on internal modeling software for doing precisely that, and IIRC our target was being within ~1% over a 10-year period (there's a bit of year-to-year variation depending on weather). Those numbers generally aren't released publicly though (because everyone wants to talk about the much bigger nameplate number), and at least as of five years ago the freely available tools for doing similar calculations were much less accurate.

I agree, there are some mitigating factors though.

One nice thing about solar in warmer climates is that its peak output is in line with peak air conditioning demands. On top of that if the solar is installed on the building roof it can lower the thermal load on the building. A double win.

Also in a ideal world we would have a world wide grid where solar energy is being routed around to where its needed with much less need for storing for night. Its always sunny somewhere right? There are of course line losses, not sure how that would stack up to battery round trip losses.

Doesn't this assume maintaining existing usage patterns? There could be enough flexibility in workloads that with an economic incentive they will be run at peak production times with only the necessities being turned on at other times.

Reply on news.ycombinator.com