Hacker News Reader

Comment by cogman10 | Hacker News Reader

If you are interested in this, you really should go watch the Tesla battery day presentation.

A few of their techniques will MASSIVELY reduce the cost per kwh of battery storage. They already have around $100/kwh tech and they are looking at reducing it to ~$50/kwh. Their big issue today is scaling. It's not that the batteries aren't cheap, it's that they simply can't make enough of them (they are scarce).

Tesla is way ahead of the curve when it comes to battery manufacturing, they have to be for their EVs. They are about to amp that up even further.

View on HN

Grid batteries have slightly different requirements compared to EV batteries. Weight/volume doesn't matter. Charge speed isn't a priority. Discharge speed isn't awfully important either. Round trip efficiency and cycle life are even more important. Temperature ranges aren't as wide.

Overall, I think these differences will lead to separate production facilities and differing designs.

According to Tesla battery day they are going to use the same production line to produce three different batteries, varying the cathode. Iron Phosphate for low-cost or full-discharge use cases, Nickel Manganese for mainstream and Nickel for high performance.

Counterintuitively it seems that "make it up with volume" is winning; rather than all sorts of different battery designs (with associated fixed costs of manufacture), they're just going to turn out a billion 18650 cells.

> Counterintuitively it seems that "make it up with volume" is winning;

The 18650 is a form factor, not the chemistry.

Tesla uses the 18650 in the Model S and X but uses a 21700 form factor for the Model 3, Powerwall and Powerpack. They announced during their Battery Day event the transition to the 4680 form factor.

The Powerwall and Powerpack uses a Lithium NMC chemistry which is different from the Lithium NCA chemistry used in their automobiles.

* Clarification

> Grid batteries have slightly different requirements compared to EV batteries. Weight/volume doesn't matter. Charge speed isn't a priority. [...]

I have the impression that Tesla tweaks the chemistry depending on the application. The cells used for power wall and grid storage might be slightly different to take advantage of those differences.

While weight/volume isn't important itself, reducing the weight/volume per kWh will also tend to reduce the cost per kWh since you're using less materials.

There are storage solutions that go further in taking advantage of the points you mention, such as flow batteries, but in that case I think it can be expensive to get high power output. The article mentions that price per kW is also important. With batteries you automatically get more watts the more watt-hours you add, which is useful in many cases.

They do, the Powerwall and Powerpack use a Lithium NMC chemistry as opposed to Lithium NCA used in their automobiles.

> Charge speed isn't a priority. Discharge speed isn't awfully important either.

Not yet, but if power production becomes more volatile (more wind/solar and ???) without associated demand response, how quickly you can charge and discharge starts impacting your duty cycle (“inventory turns per hour”).

I don't think that's necessarily the case for sufficiently large installations unless the cells are really slow. For any grid-scale battery you're going to have many thousands of individual cells linked together. Even if each cell is individually only capable of relatively small charge/discharge rates, it shouldn't matter at the scale of the whole installation for normal supply/demand scenarios. I would only expect brown-out situations if something bad happens to the grid (like PG&E decides to shut down power distribution elsewhere because of fire risk...)

I think there's some applications for high discharge cells. The local car ferry is electrifying, and they're planning on charging at every docking. On my side, our grid power is somewhat fragile, and they're adding redundant wiring between substations (it will be a loop instead of a tree), and a battery bank at the substation nearest the ferry dock, to ease the peak demand.

I think there will always be a trade-off where charge/discharge rates need to be comparable to the durations of peaks and troughs in supply/demand or else capacity must be over-provisioned. If peaks/troughs last N hours and you have K times the minimum capacity needed, then each cell needs to charge/discharge in N/K hours.

If you’re a public utility, sure.

But a faster recharge lets you do more of your buying the absolute trough. And faster discharge lets you capture more of any spike (but that’s less valuable methinks).

Kinda like being able to charge your Tesla with a bigger circuit at home so you don’t have to charge as much during peak hours.

More wind and solar tends to decrease volatility on the short term, while increasing the diurnal and seasonal variation.

A big market for grid batteries then becomes "frequency stabilisation services": they are paid not for actual turnover, but for being there and able to respond to demand within an AC cycle.

>Discharge speed isn't awfully important either

"On December 14, the Loy Yang coal power plant — one of the largest in Australia — suddenly went offline. In an instant, the grid shed 560 MW of electricity, enough to power 170,000 homes. 600 miles away, the Hornsdale Power Reserve battery system, as the Tesla system is officially known, kicked in within 140 milliseconds. It reacted so quickly, in fact, that the local grid operator was unable to measure the response time accurately. "

https://cleantechnica.com/2017/12/27/tesla-grid-storage-batt...

The article mentioned that the battery was only able to supply 16MW to the grid. By all means a significant amount, but nowhere near the required 560MW needed to stabilize the grid.

This is where hydro-based power plants excel, they're able to adjust the power output as quickly as gravity can push the water down the pipes. They're able to ramp up from almost no output to full capacity in minutes. Gas-turbines are also able to do this.

"Minutes" is not "140 miliseconds".

But it's well worth reading https://www.ofgem.gov.uk/system/files/docs/2019/09/eso_techn... on an actual UK grid trip event to see what kind of things can happen; start at "summary of event" on p4. Note the three paragraphs starting with "However" as subsequent stability systems also failed.

> This made the cumulative loss of generation 1,691MW

(!)

> All of the available “backup” power had already been deployed

That UK grid trip event was going to happen one day. The "Vector shift protection" and "Rate of Change of Frequency" in embedded generation (Home solar etc.) means that if it senses anything "funny", it'll shut off.

Which in turn means if anything else goes wrong with the grid and there isn't enough generation, then all home solar countrywide cuts out, making the problem 10x worse.

Lesson learned: "I detected something weird, lets turn off" is not a good algorithm for everyone to use on a system requiring high uptime!

It was obvious from the start really...

Reminds me of the flash crash -- "hey, the market's funny, we're all gonna pull out!"

That's switching speed, not discharge speed. Discharge speed is how many amps a battery can put out.

The amount of power for charging or discharging is not the same thing as the latency to start sending power.

A Tesla car rapidly accelerating is over a megawatt. This installation is far more cells than 560 cars worth. That means each individual cell can afford to discharge slower.

> Temperature ranges aren't as wide.

I thought people were mounting these out in the garage.

For me that would be an 80° delta between the highest summer highs and the lowest winter low. The delta in places like Wyoming would be a lot more.

Big fixed installations (think multi megawatt industrial sites) have far more batteries compared to surface area. That means it makes sense to simply wrap the batteries in insulation and use heaters to keep them warm over the winter. Sure, you use a little power to do that, but you gain back far more performance.

For $50 worth of electricity, you can keep something the size of a refrigerator 40 degrees F warmer than the outdoor air all year if you can surround it with an insulating box 2 inches thick.

In a car, the volume to surface area ratio is worse, and you can't guarantee it won't be left in the cold for months without a grid connection, so you need to ensure the chemistry can work cold.

I'm more familiar with Alaska and had a 130˚F delta between the high and low air temp. The garages I'm familiar with were always insulated and contained the furnace. Even if the garage was unheated, the excess heat from the furnace and insulation would keep the inside above freezing. Also, the insulation and lack of windows would keep the garage cool in the summer.

Presumably that’s less range than in a car driving around in Wyoming?

Yes. In fact the thinking is that increasingly grid batteries will be sodium-ion, because weight is not that big of a deal. Additionally, we have enormous quantities of sodium available, which is why it's predicted to be significantly cheaper in a decade.

Are they allowing off-grid people to buy their batteries yet? The last time I tried to talk to them, they refused to discuss anything that was not grid-tied. In the mean time, I've been looking into LifePO4 batteries as they appear to be a little safer.

Unfortunately, it looks like the their home power division is taking a back seat to their car division. I really hope that changes when they get more battery capacity.

That said, the sad truth is they may abandon it in favor of doing utility storage solutions :(. I'd really love to do a solar + power wall setup for my own home.

If you don't go through tesla, you might be able to get a powerwall supplier to install it for an off grid solution. However, I hear powerwalls are in short supply (and for good reason, they are the cheapest storage solution on the market).

> Unfortunately, it looks like the their home power division is taking a back seat to their car division.

This is not accurate. Tesla's stationary storage unit is seeing unprecedented demand outstripping their supply of cells [1], and they are prioritizing deployment in geographies where they have partnerships (South Australia [2] and Vermont's Green Mountain Power territory [3], both of which are deploying distributed storage as large virtual batteries [VPPs]). Off-grid applications simply aren't a priority at this time (and there are other vendors for such applications, such as Enphase, SunPower, and Generac, who are offering stationary storage for off grid), as grid tie batteries offer more value both to electrical consumers and the electrical grid.

[1] https://www.utilitydive.com/news/tremendous-demand-for-stati...

[2] https://reneweconomy.com.au/tesla-vpp-expands-to-add-virtual...

[3] https://pv-magazine-usa.com/2020/05/26/green-mountain-powers...

EDIT @boulos: That wasn't my interpretation, but if that's the case, I hope my citations are still helpful.

@boulos is correct. That was my bad writing :)

I don't know which has a higher priority for tesla, utility or car manufacturing. It looks like both have a higher priority than the home battery distribution (hence the ever increasing price of the powerwall without increasing capacity).

Powerwall remains the best choice in the market for home energy storage, if you can get one.

Isn’t the person you’re responding to saying exactly that? Tesla is prioritizing utility storage over home storage (and your links provide helpful evidence).

I wonder if logistics plays any role in this? Between miles traveled and not being able to use any mains power tools.

Current and last generation cordless power tools can be pretty sweet, and the ads always show professionals using them (don't you want to play dress-up like the real adults?), but I'm never clear how deep that market penetration is.

For the batteries, that's pretty terrestrial, so cords aren't that big of a problem, unless there's no power grid. I could see the solar guys being largely cordless though. I don't want to die because I tripped over your power/pneumatic lines.

(also how much of this would be solved if Tesla started using their own truck for construction?)

Agreed.

I learned about Wright's Law via Munro Live's (or someone adjacent) coverage of battery day. It's akin to Moore's Law for manufacturing, production. Paraphrasing, unit cost goes down predictably as production ramps up.

Per Munro, Tesla's projections for cost reduction nicely follows Wright's Law and is entirely reasonable.

> Their big issue today is scaling.

Brian from Real Engineering did a video on this subject recently (referencing the Tesla's battery day presentation): https://www.youtube.com/watch?v=1Xwxe0wU4b8

At $50/kwh you could battery back up a large house for ~48 hours for about $5000.

Reply on news.ycombinator.com