I'm having a hard time understanding how we can move the bulk of our energy to renewables without some kind of massive storage system. The wind can die for long periods. It can get abnormally cloudy for long periods. There's nighttime and winter.
I'm no expert, but is there any storage system that is practical that can store the amount of energy we'd need to have most (all) electricity come from renewables?
That’s why Grid Storage is a major topic in power generation and renewables. The technology and economics are evolving. A lot of people are working on finding a storage solution that is both durable and economical. One solution is probably not going to solve for all situations.
In the meantime, load sharing is used to distribute power across a power grid for cases where one area has more power and another area needs power. For instance, wind may die out in one area but over larger areas there are usually still areas with wind. It’s harder with solar, of course.
RE "....load sharing is used to distribute power across a power grid ....." TO distribute power across the grid - on a large scale fashion , would require a new power ( very expensive) power distribution network. The base problem with renewable solar and wind is its not reliable. an have outages for SEVERAL days
We can do the bulk of it, i.e. more than 50%, by geographic and modal aggregation fairly easily. Doing all of it from renewables is still an unsolved problem that would require better energy storage.
Not necessarily enough wind to be useful. If you look at graphs of wind turbine output across entire large power grids, you see 4:1 variations in a day over the PJM and CAISO regions.
That's irrelevant to this blackout, anyway. This blackout occurred during a period when supply far exceeded demand.
We have very detailed continental weather readings for over a century and can build very good statistical models. So you just ask a supercomputer to find the lowest cost solution for 99.99% electrical coverage without using any carbon sources.
It's quite possible -- over the previous 30 years we've never had an hour where there was no wind & sun anywhere in Europe. So we could theoretically solve it without storage, but that would require an insane amount of overbuild and interconnect. Imagine supplying all of Europe's electricity needs with just wind power from the south shore of Ireland.
Adding a few days worth of storage reduces the interconnect & overbuild needs dramatically, putting 99.99% carbon free within the realm of the possible.
> So you just ask a supercomputer to find the lowest cost solution for 99.99% electrical coverage without using any carbon sources.
Sure... but my expectation is that that 90% solution would include an unreasonable amount of storage (assuming you want most/all renewables). I guess was my point.
> It's quite possible -- over the previous 30 years we've never had an hour where there was no wind & sun anywhere in Europe.
But "some" wind or sun is not enough to support the whole grid.
I wouldn't assume all or even most renewables. I would assume you'd continue to use existing hydro and nuclear. The nuclear would help a little, but the hydro helps a lot. It's capacity can be saved for times when wind and solar are low.
In that scenario 90% is very easy to hit. A couple of hours worth of storage, maybe.
> But "some" wind or sun is not enough to support the whole grid.
The less there is, the more you have to over build to compensate. Doesn't make it impossible.
Having enormous synchronized grids helps a bit with that.
It’s highly unlikely to be abnormally cloudy/calm over the whole of Europe, for example, so Europe’s large grid (https://en.wikipedia.org/wiki/Continental_Europe_Synchronous...) can be used to move electricity from where there’s excess energy from solar or wind to where there’s a deficit.
That requires large capacity connections, though that aren’t everywhere yet.I understand electricity was restored in the south of France much more rapidly last week than in Spain and Portugal because it’s much better connected to the rest of Europe.
The reason why the Iberian grid is not better connected to France might have something to do with French Nuclear not wanting to be bankrupted by cheap renewables from southern countries, this was the word on the street a ehi'e back, when the push for renewables in Spain in Portugal was starting up.
This is a political problem, having Mitteleuropa depend on power from Putin is apparently OK, but on lazy southerners is verboten. Hopefully this blackout changes that (but I'm not holding my breath).
The network connection doesn't really depend on the population density around there.
I'm not saying it's a simple technical problem, but it's not insurmountable either.
Some hydroelectric plants can pump water back up and both absorb excess supply and provide stabilization during demand spikes.
I have no idea if it’s feasible to fill longer-term gaps during extended cloudy/windless days and nights using that alone, though. Other than that, there are already large-scale battery plants deployed in some cities.
Another approach is to control the demand side: When air conditioning or heating with electricity, minutes usually don’t matter, and dropping/providing extra load at very short notice should be feasible in a smart grid.
It is entirely infeasible to used pumped hydro alone to solve our energy storage problems. They’re a great component of the solution but they’re big expensive environmentally disruptive projects and there are limited suitable sites.
They're also quite inefficient (75% ish, which is not horrible, but still something) as you lose energy both in pumping and generation.
So the more you use them, the more energy you need in the first place, so you try not to use them too much. And the worst thing for a capital intensive business is to not get used much. Even with the arbitrage advantage, it takes a long time to pay it off. The grid may pay a "retainer" to sweeten the deal but you can't just build more and expect them all to get that benefit.
It's similar to the problem that "use excess energy to electrolyse water into hydrogen" has: no-one running a multi-billion electrolyser really wants to run it a few hours a day only when the grid is oversupplied. And in top of that, risk being cut off at the knees decades before breakeven if someone comes along with a cheaper/more profitable way to deal with the oversupply.
Just recently, Uber (with some partner) started testing delivery robots in my (city) neighborhood. And I love Waymo, as much as I've been able to use them. Maybe automation will change the economics.
Also, IIRC, for many years, Amazon barely squeaked a profit. They wanted to be at the low end of the margins to capture market share. Once they got big enough, they increased their margins a little and started turning a big profit.
I think uber is different though. It hasn’t been founder led for several years and the current ceo was hired to cut out all r&d spend and maximize profits and revenue.
They report things like foreign currency transactions and stock based compensation as free cash flow.
The hope with uber is that they become an aggregator of AVs which belies an assumption that autonomous vehicles will essentially be a commodity. Or perhaps that AVs take much longer to become practical at scale than people think.
But one of the biggest red flags to me with this company is how they aggressively buy back stock and publicly announce that they believe it’s undervalued. You’re in the midst of an extremely competitive emerging market and your best idea for your cash is to… buy stock?
If you are willing to put in some Project Management time yourself, you can get solar for a LOT cheaper than if you hire an outside solar company.
For the installation - most roofing companies will install the roof mounts and screw on the solar panels for $1.5k-$2k (might be different in your area, but still cheap). An electrician can be hired to make all the solar panel connections and the grid connection for a few hundred $$$.
As the Project Manager - you'll need to source the panels, file the permits, and hire the roofers and electricians. https://www.solarwholesale.com/jerry-rig-everything/ can assist (assist, not do) all of that for you for a cheap price. $10k-$15k with Solar Wholesale will get you the panels and cables you need, plus the engineering drawings, the permit paperwork, and etc. (thank you for the great Solar DIY series Jerry!)
For $20k and working as the Project Manager yourself, you can get what most solar companies will charge you $50k for. Which is ballpark inline with construction project management costs - typical residential construction project management cost 40-50% of the total project budget. A lot of those projects as complex enough to warrant those fees, unless you have a lot of experience, but in my opinion adding solar to your roof is not a complex process and does not warrant those fees.
I went through this process myself as well. A few conclusions I made:
1. The installer you choose is more important than anything else. The panels are pretty commoditized at this point. The battery cells are pretty much commoditized as well (they are 2-3 companies all of the major battery companies use). Your biggest worry is not if, but when, things break. Most well known manufacturers (REC, Enphase, Canadian Solar, Hyundai, etc.) have good warranty policies, but you need someone to coordinate that - so make sure your solar installer does that on your behalf.
2. A Powerwall battery is 13.5 kWh and has a 11 kW inverter. The base is $8k and $6k for each battery expansion. A similar, well known, brand EG4 sells the same battery for $3.7k and a similarly spec'd inverter is around $2.5k. Generally speaking Tesla installers quote much higher on labor than others too.
3. From the installers I spoke to they dont really like working with Tesla.
4. If you live in a place like Texas, Florida, etc., I wouldn't even bother with batteries unless you have a relatively small house. Most houses suck down so much energy that a natural gas generator is way more cost effective.
Friend ordered the solar roof (the solar panels that look like shingles), and after waiting a year switched to their regular solar.
From my conversations it was a complete mess to get installed, with poor communication on their end. After it was installed he had 2 issues within 2 years resulting in them needing to replace hardware.
Add in that it was quite expensive, he is not a happy customer.
I went through this process this year. Found solar installation companies on Yelp, got some quotes, chose an installer. A month later, I have Panasonic panels and a Powerwall powering my home. There’s plenty of info online. I live in the US and found this resource very educational: https://www.energy.gov/eere/solar/homeowners-guide-going-sol...
We have a Tesla car, but I never considered their solar because it's a commodity, and their options are overpriced. They also pushed leases pretty heavily, which I was not interested in. We bought our panels in 2018, the same year we bought the EV.
Just talk to your neighbors about reputable local installers. They'll have panel and storage options.
That said, because these things are largely commoditized, there's some turmoil in the market. Our (excellent!) local installer was saved by the PPP during the pandemic, which was a bit of a sobering thought -- they're the ones backingthe 25 year installation guarantee!
Then Sunpower (our panel vendor, and one of the higher end ones, at that) went out of business. Our monitoring system continues to work thanks to the work of the bankruptcy courts, though it's been transferred to a company who is trying to support those assets be having a freemium model.
Except you can guarantee MKBHD got the best of the best service with Elon Musk personally holding a gun to the head of whomever was in charge of the install.
They literally work with their top influencers for FSD to make it looks more impressive than it really is.
He even talks about how everyone was having problems but "for whatever reason it went perfectly"... and it still took 8 months
You should look closely at solar rate plans and interconnection fees in your area. If you're hoping to save money with solar, it's possible that you never recoup the initial investment because of the way the newer rate plans are structured (esp. for PG&E).
In the bay area, it might be cheaper to opt into 100% renewable rate plans from CleanPowerSF or Peninsula Clean Energy. (Silicon Valley Power is 100% renewable by default, IIRC.)
Not political => without even looking at the solar panels, I'd be wary based off all the numerous issues with their cars. The CyberTruck in particular is like a modern day DeLorean, but much worse. That kind of cost cutting is going to be company wide and makes me wonder if their solar stuff is also problematic (not that I've heard anything).
Political=> he did a full blown Nazi salute twice and has been working to have millions fired from their government jobs. The conservatives who don't seem as bothered by that have never been interested in EVs to begin with and his liberal fanboy customers will never buy from him again now. He is extremely unpopular in Germany and China's Tesla market is being out competed. I don't think that makes Tesla look good long term, so I'd be cautious about a long term investment with them like rooftop solar. That's just my opinion anyway.
I will never buy another Tesla again for political reasons, but regarding reliability: their new models have always had reliability problems, but then reliability has always gotten much better within a year or so.
I don't know if the Cybertruck will follow the same pattern, or if the whole company has jumped the shark, but if we're looking for non-political opinions I would not necessarily write them off on quality issues alone.
I'm sure that happens a lot. In my experience, people tend to be bad at estimating things.
As a long-time SWE who eventually (also) managed teams: I found estimates particularly difficult. Some SWEs would give very optimistic estimates (and always miss). Others would "sandbag" their estimates, because they didn't want to have to explain why they were late.
I tried really hard to make sure the SWEs on my team didn't get "blamed" if they were late and asked them to generally give aggressive estimates. My rationale was that if you estimated more time than it really took, you would tend to fill up that time no matter what.
It depends what you do with the estimate. If you have to deliver a product/a functionality where X people depends on it to continue. It's best, in the general case, to overestimate a little to not have the X people preparing to work when they can't. If you estimate to schedule the task for the week, it can be more useful to be more aggressive since missing the deadline is less impactful.
For example, there are a lot of task that I depends on which can be completed in one week or three month. I don't care which it will be, I just want to know when it will be done so I can schedule other tasks before.
> Some SWEs would give very optimistic estimates (and always miss).
This seems to be the easiest case. Keep a spreadsheet with two columns: the time they estimated, and the actual time. Calculate the average coefficient between the columns. Next time they give you an estimate, don't comment on that, just privately multiply it by the coefficient.
Nvidia in same timeframe far outgrew Meta, maybe no other major ones, except maybe netflix recovering from an even earlier drop. A good portion of Meta's gain was just getting back to where they were before the Apple store changes induced drop (that also hit Twitter).
I'm wondering how small of a model can be "generally intelligent" (as in LLM intelligent, not AGI). Like there must be a size too small to hold "all the information" in.
And I also wonder at what point we'll see specialized small models. Like if I want help coding, it's probably ok if the model doesn't know who directed "Jaws". I suspect that is the future: many small, specialized models.
But maybe training compute will just get to the point where we can run a full-featured model on our desktop (or phone)?
> Like there must be a size too small to hold "all the information" in.
We're already there. If you running a Mistral-Large-2411 and Mistral-Small-2409 locally, you'll find the larger model is able to recall more specific details about works of fiction. And Deepseek-R1 is aware of a lot more.
Then you ask one of the Qwen2.5 coding models, and they won't even be aware of it, because they're:
> small, specialized models.
> But maybe training compute will just get to the point where we can run a full-featured model on our desktop (or phone)?
Training time compute won't allow the model to do anything out of distribution. You can test this yourself if you run one of the "R1 Distill" models. Eg. If you run the Qwen R1 distill and ask it about niche fiction, no matter how long you let it <think> for, it can't tell you something the original Qwen didn't know.
I suppose we could eventually get to a super-MoE architecture. Models are limited to 4-16GB in size, but you could have hundreds of models on various topics. Load from storage to RAM and unload as needed. Should be able to load up any 4-16GB model in a few seconds. Maybe as well as a 4GB "Resident LLM" that is always ready to figure out which expert to load.
> We're already there. If you running a Mistral-Large-2411 and Mistral-Small-2409 locally, you'll find the larger model is able to recall more specific details about works of fiction.
Oh, for sure. I guess what I'm wondering is if we know the Small model (in this case) is too small -- or if we just haven't figured out how to train well enough?
Like, have we hit the limit already -- or, in (say) a year, will the Small model be able to recall everything the Big model does (say, as of today)?
It's a sliding scale based on what you consider "generally intelligent" but they're getting smaller and smaller. This 27b model is comparable to 400b models not much over a year ago. But we'll start to see limits on how far that can go, maybe soon.
You can try different sizes of gemma3 models, though. The biggest one can answer a lot of things factually, while the smallest one is a hilarious hallucination factory, and the others are different levels in between.
> It comes in sizes from 1 billion to 27 billion parameters (to be precise: 1B, 4B, 12B, 27B), with the 27B version notably competing with much larger models, such as those with 400B or 600B parameters by LLama and DeepSeek.
> I've gone back to pirating everything. I can afford to pay for all the services, it's just the service and content quality has gotten so bad that it's just not worth it.
If the content is so bad, then why the need to pirate it?
I'm no expert, but is there any storage system that is practical that can store the amount of energy we'd need to have most (all) electricity come from renewables?
reply