Hacker News Reader

Comment by matt2000 | Hacker News Reader

matt2000

The decreases in battery cost in this report are pretty spectacular, but it's missing comparisons to costs for traditional generation. As best as I can tell from https://www.eia.gov/outlooks/aeo/pdf/electricity_generation.... - the LCOE for gas generation is around $40/MWh. So this article's cost of $187/MWh for batteries is still a lot higher, but the battery cost has dropped 35% in one year so it could get there soon. What I'm still not sure about is if you need to include the LCOE of the original generation method in the cost of the battery system? In that case the total LCOE for say wind+battery would be ~$210.

View on HN

$40/MWh is for combined cycle gas turbines that operate at high capacity factors. According to this BNEF report, gas peaking capacity is now threatened by battery replacement:

Electricity demand is subject to pronounced peaks and lows inter-day. Meeting the peaks has previously been the preserve of technologies such as open-cycle gas turbines and gas reciprocating engines, but these are now facing competition from batteries with anything from one to four hours of energy storage, according to the report.

The report itself doesn't give a $/MWh figure for these peakers. Lazard's 2017 report puts the lower end of CCGT generation at $42/MWh, close to the EIA number, but gas peaking starts at $156/MWh and goes as high as $210/MWh:

https://www.lazard.com/media/450337/lazard-levelized-cost-of...

Note that they put gas reciprocating engines no higher than $106, so I don't think that batteries at this price threaten gas reciprocating engines yet. Mostly they threaten open cycle gas turbines.

Diesel reciprocating engines show a cost of at least $197/MWh and are also threatened by battery-backed renewables. Diesel generators have been heavily used to supply electricity for small remote villages, islands, and off-grid mining sites. For a few years now there has been a trend to reduce consumption of diesel at such sites by partially substituting generator output with renewable electricity. It's possible to make deeper cuts in diesel use with added battery storage, and the payback period is shorter than you might guess from looking at the local gas station's diesel price. Getting the fuel to certain locations can cost nearly as much as buying it in the first place.

This is the thing that gets missed is renuables are competing with peaking power which is more expensive than base load power. The way to think of baseload power is it's the lowest quality power source. If you contract for it you have to use it. And it's generated when other consumers don't need it. Which is why it's cheap.

Batteries are competing with very expensive grid stabilization technologies. Solar is selling into the peek daily power cycle.

If batteries are the enabler, couldn't you just over generate using a cheap base load tech, dumping excess to battery? Or are renewables plus batteries cheaper than base load tech plus batteries?

Yeah battery storage is a arbitrage business. Buy cheap sell high. Buy midnight wind from North Texas at three cents a kwh and resell it on a cloudy humid afternoon in Phoenix for thirty.

Currently fossil fuel and nuclear power is really cheap late at night. In the future it'll be less predictable but there is absolutely a business there.

> This is the thing that gets missed is renuables are competing with peaking power

The other thing that gets missed is that they aren't competing at all, because renewables are unreliable and non-dispatchable. So, more renewables means more peaking power other things being equal. What can directly compete with peaking power is battery storage (though pumped hydro is king there) and demand-response.

Renewable + storage are competing directly with peakers. The storage makes renewables dispatchable and if you look at the performance of the battery in Australia it appears to be much higher performance. The system dispatchs power in a fraction of the time a peaker plant can.

Takes a lot less time to throw a switch on what is essentially a capacitor than start a gas turbine, even if its the small peaker gas turbines.

Wind and solar are baseload power. Very cheap, but don’t follow the demand curve. That’s a fairly direct replacement for coal and nuclear, it’s just peeking power in we need to deal with.

Because they are so cheap wasting some power output each day is simply not as big a deal. The balance point between extra wasted production, storage, and peaking power plants is not obvious or nessisarily stable as prices change.

Wind and solar are neither baseload nor peaking power sources. They are intermittent sources where you need other parts of the grid to smooth out the intermittent nature. Either via storage, via peakers or via additional baseload on top and something that consumes the excess that intermittent + baseload can produce.

Sun is totally out during the night. And wind is only as much a reliable baseload provider as it will provide during the worst days of the year, which in many locations can be a tiny fraction of their average load factor.

Many good points above, I just wanted to point out that there are lots of places like where I live, where peak electricity is nearly completely correlated with the sun shining (air conditioning). Not that this matters for places where it's not that way, just pointing that out.

The fact that wind and solar don’t produce steady state power is completely irrelevant.

What power companies care about is the difference between the cheap energy sources production and demand. Night time demand is often so much lower than peak demand that it takes less peaking power plants to cover solars night time deficit than coals daytime deficit.

PS: The difference between winds minimum expected output and average output is also smaller than most assume. Locations that get more wind at specific times of the day are common and let you tailor supply and demand. Unusually high winds end up wasted, but discarding 5% of output does little to change the relative costs.

> The difference between winds minimum expected output and average output is also smaller than most assume.

https://i.imgur.com/UTjUAp4.png

Expected output is the important point.

https://carboncounter.files.wordpress.com/2015/08/wind_hourl...

Wind over large scales is less random than people assume even if the variation is large. California both needs more electricity in the summer and gets more wind energy in the summer. Further, peak solar and peak wind output occur at different times of the day which again is extremely useful.

If the actually observed output were lower than the expected output then those expectations require adjustments. And germany is about as large as california. It also includes offshore wind, which california does not even have.

Maybe your weather is more stable and does not generalize well.

I'm not sure if these numbers being quoted are good for analysis.

The "Nessie Curve" shows that power at 5:00pm is worth more, especially in sunny environments (like Hawaii). Solar energy is taking over those areas, but solar power begins to drop dramatically as the sun sets.

5:00pm to 8:00pm is still quite warm, so you need to turn on the gas turbines to provide electricity. But solar's efficiency has dropped dramatically, so you can't really rely upon solar power in those hours (well... you can... but at dramatically lowered efficiency).

I'm not sure any analysis is worthwhile unless it includes the time-of-day, as well as the number of hours that the batteries can load-shift power. 5:00pm to 8:00pm power is going to cost more in the future than 12:00pm power, simply due to this whole solar energy thing going on.

I just wanted to make clear that BNEF is talking about batteries threatening fossil peakers that produce expensive electricity, while OP was wondering how inexpensive CCGT generation could be threatened by batteries. CCGT and peakers are distinct and have different costs even when they both burn natural gas.

Hawaii doesn't generate electricity from natural gas but it does consume a lot of petroleum-fueled electricity:

https://www.eia.gov/state/?sid=HI#tabs-4

Here's a story about a Hawaiian island replacing diesel output with solar and batteries from a couple of years ago:

https://arstechnica.com/information-technology/2017/03/kauai...

More recently, the Hawaiian Public Utilities Commission has approved 247 megawatts of new solar capacity backed by nearly a gigawatt-hour of battery storage:

https://pv-magazine-usa.com/2019/03/28/hawaiis-new-reality-o...

"The price for each of these contracts was between eight and ten cents per kilowatt-hour. This is cheaper than both gas peaker plants and HEI’s current cost of fossil fuel generation, much of which is petroleum-based, which the company put at around 15 cents per kilowatt-hour."

Note that these prices are only $80-$100 per MWh, because most of the solar electricity is consumed immediately and doesn't need to be stored in a battery.

I'm positive Hawaii could use geothermal power for base load. All those volcanoes. Iceland and the Azores are already doing so.

> I'm not sure any analysis is worthwhile unless it includes the time-of-day, as well as the number of hours that the batteries can load-shift power. 5:00pm to 8:00pm power is going to cost more in the future than 12:00pm power, simply due to this whole solar energy thing going on.

A big part of the problem talking about these things is that people conflate photo-voltaic with solar thermal (CSP).

What you're talking about when you say 'solar' is just the first one.

Solar thermal plants provide power into the night.

Arguments that it's just solar + storage are perhaps valid, but a) it's still solar, and b) nuclear MSR's aren't called 'nuclear + storage' (ditto anything else using latent heat in fluids, flywheels, etc).

It does seem that Solar Thermal plants are better for the expected future of energy demand. I wasn't thinking of them for some reason.

But yes, people need to be mostly aware about the time-cost of energy. Even if Solar Thermal is less efficient than photo-voltaic cells, the fact of the matter is that 5:00pm to 9:00pm power is the REAL problem that people need to focus on. That's when America uses most of its electricity right now, and is likely the main driver of peaker plants at the moment.

People ignore CSP all the time. I estimate that by 2050, when most of the grid electrical capacity is coming from MS storage with operating delta_K's around 1000+, they'll learn.

I think one thing that needs to be considered is for HVAC you can time shift your demand via thermal lag. So for air-conditioning you can pre-cool the house during the day and just let the temperature rise during the evening.

Would people live with that? They would if doing that saved them $250/month.

There are already ice-based thermal storage systems exactly based around shifting the electricity consumption for cooling to off-peak times:

https://www.ice-energy.com/technology/

So you don't even have to let the temperature rise during the peak-demand period.

A friend in Arizona mentioned having solar panels and setting the thermostat to 65 degrees between 10am and 5pm. And off after that. Later at night they open the windows and close them in the morning.

They're just using the thermal mass of the house to time shift energy.

The feasibility of this is completely determined by the insulation and albedo of the house. Working around the thermal inertia of the house is tricky, blasting the AC hard at night is only effective if your house is moderately well insulated or good at not absorbing heat (reflective roof, wall insulation, reflective blinds etc, and by reflective I don't mean tinfoil, I mean like white or bright colored, basically no black or dark). Building elements are remarkably bad at storing heat/coolth since their sensible heat capacity is quite low. There are ways to enhance this effect though by getting building elements that are doped with phase change materials.

Of course the insulation part of the equation is a double-edged sword. You want insulation, but you always want air circulation at a rate conducive to low CO2 levels. Maybe a compromise using something like Sheetrock with extra paraffin, and the whole thing painted Anti-flash white, with air exchange passed through heat exchangers underground.

It would be pretty expensive to maintain though.

In modern homes air tightness and mechanical ventilation are both required by code in many places to deal with the issues you've raised.

Even a crappy insulated house has some thermal lag. Which is what you are playing with.

I think people get confused because a low carbon grid is going to have a different pricing structure than the current one where 'base load' power is cheap at night. The reverse will be true. Power in the evening is going to be spensive. With the cheapest power at noon.

The solution is to time shift demand. A lot of demand can be time shifted. HVAC can be time shifted using thermal lag and storage. You don't really need batteries for that.

Solar is most effective at 12-13pm due to panel orientation and noon insolation. At 5pm it's sonething like 50-60% of noon insolation during the summer (best case scenario).

>> number of hours that the batteries can load-shift power

Isn't that defined by the price per MAh of the batteries which was mentioned in some of the parallel threads?

Note that diesel generators have the counter intuitive property of using more fuel at 50% of max load than at 80% of max load. Most other power generations have some variation of this where operating a 80-90% of max power is where they are more fuel efficient. Thus even for conventional generation battery backup on the grid has the potential to save money.

In diesel engines the power generated is directly controlled by the amount of fuel delivered - if you want more power, inject more fuel. Unlike most spark ignition engines, there is no throttle valve, you control the stroke of the diesel pump (or the time the injectors stay open in common rail engines). So it is completely wrong to say that they use more fuel at 50% load than at 80%.

citation needed. I could see an engine using more fuel per unit power at a lower power level but not less fuel overall.

If they did you would just run the engine at 80% all the time and waste the excess in a massive water brake or resistor bank.

I am sure it is per unit. Means you are better off to run it at high load and battery the excess power, then shut the whole thing off for a bit, vs run it at 50% all the time.

If you have two engines you could be running at 40% each, or shut one down and run the other at 80%, it will be more efficient to do the latter (unless you have to be able to start the idle one quickly and that's not possible if it's at 0%, so then you run one at 75% and the other at 5%).

Engines are simple pumps that operate most efficiently “fully open”

I don't know about using more fuel at 50%...

They are however, are more efficient around 70% load than 50% load.

And that’s why Kauai is starting to replace their diesel generators with solar + Tesla batteries.

LCOE includes all the costs, including the power to charge the batteries (assumed to be 60% of wholesale power price, since you can charge whenever it's cheapest in the daily cycle.)

I may be wrong in the interpretation, but LCOE would seem to apply only to the actual capacity, and not the grid‘s entire throughput.

Yet batteries would only be needed for a fraction of the installed power, to adjust for fluctuations in supply and demand. Most power would never see the battery in such a grid.

And, of course, batteries are only one of multiple strategies to get supply and demand to equalize. Smart consumers is probably the most underdeveloped now, from cars that would charging (and possibly even discharge into the grid if your schedule allows), washing machines picking the best times when possible, or cooling and heating working with their respective reservoirs.

Batteries have cycle life of thousands of cycles so you can't really compare the costs of production and storage directly.

I'm sure his numbers take that into account. The cost per MWh for batteries would be at least an order of mangnitude higher if he did not.

You are correct. Capital cost of battery systems is closer to $800/MWh

Page 11: https://www.lazard.com/media/450774/lazards-levelized-cost-o...

I'm curious how they come up with that, actually. Aren't batteries at ~100/KWh at the moment? I would have expected a MWh to be closer to $100k. If I'm not mistaken, 1000 cycles at that rate would get you close to $100/MWh.

Obviously I am misunderstanding something in these calculations.

I assume that one difference is that a facility for LiIon batteries costs more than just LiIon battery cells; and the land, the transformers, the site security, and general maintenance of all the above are non-trivial compared to cell costs?

Yup. Capital cost, operation and maintenance, and charging costs are all added up, financed over the operating period, and turned into a cost to meet an investment IRR of 12%. The table in page 11 of the pdf I linked above breaks it all down.

12% seems massively high for the risk level. This should be extremely low risk, right?

I think you're reading it as Annual Recurring Revenue rather than Internal Rate of Return. For IRR I think 12% is pretty normal, and depending on how good your project management is it could easily turn into 0%. But I'm not an accountant so I might be misunderstanding.

Investing in a single facility with unproven technology, without other existing facilities with well known financials is a very risky investment.

Oh certainly, but that is kind of my point. The $/KWh seems way too low if that is installed cost. But the levelized cost of energy at the end actually seems pretty reasonable.

That means I don't understand where the $/KWh of installed capacity comes from.

Sure. I don't know what I was thinking.

Batteries don't generate power so comparing them to power generators is like comparing apples to apple trees.

Reply on news.ycombinator.com