Hacker News Reader

Comment by epistasis | Hacker News Reader

That post makes me distrust all the rest of his analysis. This conclusion in particular:

> Rather, the lesson is that we must work within serious constraints to meet future demands.

Within just a few years, his spherical cow estimates of needs have been proven to not be very useful for scoping the problem. And his proposed solution of nuclear has proven to be infeasible and too expensive.

So what pretends to be an unbiased assessment based on physical principles is revealed to actually be a huge number of assumptions that are not reflective of reality, or useful for thinking about the future.

This is the exact problem that the original post talks about. We are too easily fooled by models that are simple, and wrong, like what dothemath presents.

View on HN

You're not alone. That article was the reason I stopped considering him a reasonable source.

>I’ll use lead-acid batteries as a baseline. Why? Because lead-acid batteries are the cheapest way to store electricity today.

That wasn't even true when the article was written. Pumped hydro has always been cheaper than batteries, and unless we've made some big improvements since I last checked the news, it still is.

>And lead is a common element, being the endpoint of the alpha-decay chain of heavy elements like uranium and thorium.

What's really incredible is that a physicist would make this argument. Lead is not common, and anyone with even a superficial familiarity with the process of stellar nucleosynthesis can easily explain why. A zinc nucleus needs to capture around 150 neutrons to produce a lead nucleus. All heavy elements are rare on a cosmic scale:

https://en.wikipedia.org/wiki/Abundance_of_the_chemical_elem...

But the other glaring hole in the analysis is the lack of reference to prior work. Japan had already developed a grid energy storage system based on sodium-sulfur batteries in the 1980s [1]. I would expect a serious analysis to consider the existing state of the art.

These mistakes don't strike me as arising from a lack of competence, but rather from a desire to inflate the apparent strength of the conclusions.

1: https://ceramics.onlinelibrary.wiley.com/doi/abs/10.1111/j.1...

Is pumped hydro really able to scale to a significant level where it can provide enough energy storage for a national (or even state/regional) grid? I was always under the impression that pumped hydro was indeed awesome, very cost-effective, etc. but could only be developed in places with very specific geographies. After all, you have to somehow place two reservoirs near each other with a significant elevation difference. Isn't this the limit that governs pumped hydro as a storage technology?

>Isn't this the limit that governs pumped hydro as a storage technology?

I was nitpicking there, yes. Pumped hydro has an "asking if we [could/should]?" problem: extensive use of pumped hydro would be devastating to ecosystems. There are a number of "clever" strategies, such as allowing the lower reservoir to be the ocean:

https://municipalwaterleader.com/implementing-oceanuss-pumpe...

but I did not intend for that sentence to be read as advocating widespread uptake of pumped hydro. It's convenient in certain places, and it can fill in gaps for communities in need, but it comes with a big cost not measured in dollars. And the use of saltwater makes this problem much worse.

Tom Murphy covered this in the series as well. While there are absolutely cases where pumped hydro is extremely effective (e.g., Dinorwig Power Station), the inherent problem is that it requires need very specific geological features. Unsurprisingly, we've hit much of the good low-hanging fruit, so future projects will be less efficient in terms of capacity per dollar invested. And also unsurprisingly, these features are not evenly geographically spaced around the globe.

https://dothemath.ucsd.edu/2011/11/pump-up-the-storage/

I believe you're ignoring the overall point of the article, which is not that lead-acid batteries specifically are unworkable.

> Rather, the lesson is that we must work within serious constraints to meet future demands. We can’t just scale up the current go-to solution for renewable energy storage—we are yet again fresh out of silver bullet solutions. More generally, large scale energy storage is not a solved problem. We should be careful not to trivialize the problem, which tends to reduce the imperative to work like mad on establishing adequate capabilities in time (requires decades of fore-thought and planning).

He further goes on to discuss gravitational storage (e.g., hydroelectric dams and pumped storage), kinetic storage (e.g., flywheels), spring storage (e.g., compressed air), and chemical storage (e.g., batteries, fuel cells).

Again, the point is:

> With the exception of the feeble gravitational storage example, each of the ideas presented here are technically challenging, expensive, and sometimes dangerous.

And further, to contrast them to the miraculous gift that fossil fuels have been:

> A short digression to contrast the miraculous energy density in fossil fuels: our 3 days of electricity storage at 30 kWh/day requires just 12 gallons of gasoline (1.6 cubic feet; 45 liters) burned in a 20% efficient generator (it seems like the other 80% is noise!). The Earth’s battery—a one-time gift to us—turns out to be vastly superior to any of these other “solutions” in terms of energy density and long-term storage, measured in millions of years. It will be sorely missed when it’s gone.

https://dothemath.ucsd.edu/2011/09/got-storage-how-hard-can-...

> I believe you're ignoring the overall point of the article, which is not that lead-acid batteries specifically are unworkable.

Right, the point of the article is that storage is unworkable, and lead-acid batteries are used therein as a straw-man. You underestimate how much time I've spent studying this, and how many times I've read that absolutely infuriating article.

>He further goes on to discuss gravitational storage (e.g., hydroelectric dams and pumped storage), kinetic storage (e.g., flywheels), spring storage (e.g., compressed air), and chemical storage (e.g., batteries, fuel cells).

But he does not discuss the most significant existing application of batteries for grid storage. So when he says this:

>We can’t just scale up the current go-to solution for renewable energy storage

He hasn't even considered it! Granted, Na-S currently lags way behind Li-anything in costs, but that's a result, mostly, of innovation aimed at cars.

I would be interested in seeing your calculations, but so far every time I've seen someone criticize the article you're referring to they refuse to do any calculations to support their argument. Or even to show where his calculations are erroneous.

If storage is as workable as you say then it should be easy to demonstrate it mathematically.

"For every complex problem there is an answer that is clear, simple, and wrong." The argument is that these calculations are not relevant. That's the entire point of the original article that we are all posting under. Further, it's not even clear what calculation you want others to make. What's the question?

Malthus made a correct calculation. But it was irrelevant, and led to a bad correction.

There's a difference between what we calculate and what we want to know. And when that "calculation" is limited to simple napkin math, it's even less likely to be relevant or interesting to the real world.

Taking simple physical limits and deriving calculations about the world has yet to offer insight about technological change. Perhaps one could do that to calculate a potential limit to Moore's law, by saying here's a minimum size of a transistor, and we are reaching thay. But what about going to more layers than we currently put in transistors, instead of simply shrinking the transistor size to increase transistor count in an IC?

The calculation that actually matters is the amount of storage needed. Unfortunately, the article is extremely hand wavy about that part and so even though the cost estimate of storing a entire week’s energy consumption might not be far off, the conclusions drawn from it might be totally wrong

> Within just a few years, his spherical cow estimates of needs have been proven to not be very useful for scoping the problem. And his proposed solution of nuclear has proven to be infeasible and too expensive.

Care to elaborate?

Not the one you’re responding to, but I can take a stab:

The article on battery capacity assumes a country would need to store an entire week of energy to have a dependable grid based on renewables. But it isn’t like the sun just might not rise a few days in a row (and solar panels produce a non-trivial amount even on cloudy days), and it is basically guaranteed to always be windy somewhere. Overbuilding generation capacity is also missing despite being a far more economical approach to dealing with long stretches of reduced wind/solar output.

The net result is he overestimates the needed capacity by probably 1-2 orders of magnitude.

As a smaller point, his math works out to $74/kWh (in 2012 dollars) which probably seemed outrageously low at the time, but thanks to Li-Ion tech and lots of investment might actually be reached in not too long

If you've done a calculation like his I'm interested in reading it. It is hard to evaluate your argument against his since you've made no quantifiable claims.

My point is the same as the original article. Calculations like this, of single routes, are pointless and mislead rather than inform. But it's an intellectual honeypot, because it seems interesting.

The paths of possible technology are a huge high dimensional space, but let's think of simplify it to a map of geographic space. He's taking out a telescope, pointing in a single direction, and sees a Cliff really really far off, and says "well I guess there are physical limits!" Which of course. But that's not interesting, what's interesting are which path are out there, and to explore that you have to point the telescope in lots of directions, or even better yet, start exploring territory by moving around. It might be that there's a hikable path right next to the cliff that you didn't see because of the narrow view of the telescope.

And those alternate paths are what the original article is all about. We didn't run out of food. Technology changes, and we become far more efficient and productive. And pretending that there's a physical limit somewhere without bothering to peak around is a classic way that we trick ourselves about the future.

> My point is the same as the original article. Calculations like this, of single routes, are pointless and mislead rather than inform. But it's an intellectual honeypot, because it seems interesting.

He doesn't calculate single narrow routes. In fact that's almost entirely opposite the purpose of his articles, which is to take a step back and look at things from a very broad perspective: what's the scale of our energy use, what's the scale needed to replace it with something else, and what are some back-of-the-envelope calculations we can do to get an intuitive grasp of the problem?

It's essentially applying fermi estimation to the problem, which I think most people would agree is far from what you're accusing.

> We didn't run out of food.

This was never claimed?

> Technology changes, and we become far more efficient and productive.

This is addressed, particularly in the second article linked, which itself is a highly-summarized form of his entire position.

> And pretending that there's a physical limit somewhere without bothering to peak around is a classic way that we trick ourselves about the future.

There's no pretending. There are real physical and thermodynamic limits that physicists currently know no way to circumvent, and that we have increasingly convincing reasons to believe are fundamental. Pretending these don't exist is a classic way that we trick ourselves about the future.

> There are real physical and thermodynamic limits that physicists currently know no way to circumvent, and that we have increasingly convincing reasons to believe are fundamental

Name one of these physical limits that pertains to the problem at hand.

Of course there are limits. That's not interesting. What would be interesting is a limit that imposes a course on our technological transformation. Lead ain't it.

Reply on news.ycombinator.com