Hacker News Reader

Disclaimer: I helped create those Carbon Tracker graphs.

> - The author states that renewables are on the verge of being "cheaper than the cost of continuing to operate existing coal- or gas-fueled power plants". They support this argument with some cherry-picked examples. Realistically, that is not true, nor will it likely every be true.

In the case of coal, it certainly is true in many regions. Natural gas I agree is quite different, particularly given the role it will play providing flexible capacity as renewables penetrate further. But across Europe new renewables are cheaper to build than to operate existing coal, due to ageing fleets, tightening air pollution regulations and the carbon price. Same for the US, except without the carbon price. [*edit - And obviously the rapid cost reductions in renewables!]. And we see quite strong trends for Asia, SEA etc.

You can find more detail on the coal trajectories in our global report, Powering Down Coal and online portal [0].

[0] https://www.carbontracker.org/reports/coal-portal/

Apologies :).

However, you really cannot compare coal to renewables directly, since it's an apples to oranges type comparison. If you include externalities like a carbon tax in the cost curves then the you really need to understand the nuances to properly interpret those graphs.

I should have said please keep the nuances in mind when trying to interpret the graphs, and thank you for your efforts!

And are you including subsidies when you plot your cost of solar?

are there even subsidies left on solar on commercial installations?

No worries at all!

Regarding the technology comparison, that is true from a power system point of view - but for a utility which is making investment decisions in new generation capacity, it isn't so different. We include carbon taxes because coal power operators suffer that tax, it isn't an assumption about future policies. And equally to someone else's question about whether subsidies are included in solar costs, the answer is no. Because the LCOE is calculated from current module, balance of system and soft costs, and that while subsidies have brought the costs down, they aren't a component of the LCOE. As for system costs which perhaps you are referring to, that is fair from the system viewpoint but not particularly for the marginal unit of new capacity. UKERC have done some good research in the UK on system costs of renewables integration which perhaps you are referring to [0].

But you're absolutely right that there is a lot of nuance, especially around regional power market differences.

[0] https://www.carbonbrief.org/in-depth-whole-system-costs-rene...

> but for a utility which is making investment decisions in new generation capacity, it isn't so different.

You are probably familiar with all of this, but that's a tricky one since dispatchability puts those into different equivalence classes. California ISO has negative LMPs right now, which means building additional solar provides little marginal value to the utility. As a result, even if the "traditional" LCOE for solar (just using capital cost, O&M, and capacity factor) is lower than coal, coal would technically still have higher value to the utility.

You'd have to model real-time demand and fratricide to get a average marginal rate (almost impossible to predict for a 20-30 year horizon right now, given rate of innovation), and then compare that against the amortized capital cost and O&M to get an expected ROI for the plant. An LCOE comparison wouldn't really make sense from a decision making standpoint.

That wouldn't be true if they were both closely equivalent (say NGCC vs NGCT, or coal vs nuclear). Then a strict LCOE comparison would be useful.

Fair point about California's negative LMPs, but I think for a lot of regions in the US there is still a lot of room for renewable growth.

But it is fair to compare LCOEs because solar usually gets its value through long-term PPAs; either utility to generator, or even through the rate-base, and the price of that is effectively set by the LCOE. Despite missing out a lot of the other factors that you reasonably bring up, from system costs to locational factors.

With some personal interest and research I've judged for two decades, (before the IPCC managed to reach consensus) that the potential economic performance of wind and solar is constantly under-estimated and this belief, (admittedly partly intuitive) has year after year been proven true. Even while lacking investment programs of trillions of dollars into R&D which would have been entirely sensible and affordable, proven by greater amounts spent on conflict in the name of security.

I find it beyond justification or excuse how major official forecasters can year after year be so vision-less and pessimistic. We can make floating deep-sea wind farms - and floating solar. We can make better and better batteries and storage facilities as soon as dispatchability and curtailment are close to being a significant cost in a handful of places. Cost of renewable energy has cruised downward with tepid incentives provided by governments. It can plummet within years of serious international public investment in R&D.

There's really no excuse for persistently pessimistic predictions like this [1], from major and professional industry forecasters -

[1] http://www.mining.com/web/charts-energy-experts-hilariously-...

> With a ~20% learning curve, that is ~5 doublings from today, so more than net electricity demand

Just to pick up on that point, electricity demand will have to be a lot higher in future, because we're going to need to electrify transport and heating. That would be worth perhaps another doubling or more in electricity demand. And road transport at least looks like it's locked in to happen over the course of 2-3 decades through pure market forces.

Global electricity demand will skyrocket as places that now have less than ideal grids come online and local proliferation of cheaper electronics continues.

"proliferation of cheaper electronics"

Electronics use very little electricity.

The biggest users of electricity are things with motors or heaters. Washing machines, dryers, ovens, kettles, etc. Electronics by comparison are a rounding error.

Except for places where there are a LOT of electronics, like datacenters.

Even there - the lions share of electricity is consumed COOLING those electronics, but it is a fair point.

That is not correct, cooling uses maybe 20% of the electricity used by the electronics themselves.

But aren't all electronics just converting 99% of electricity to heat? Electronics don't just consume electricity by themselves -- they convert it to heat/light or noise.

True, but an air conditioning system removes multiple kWhrs of heat for each kWhr of electricity.

See for example here: http://www.currentforce.com.au/coefficient-of-performance/

As I understand it, cooling the data centres uses as much electricity as the servers themselves use[1].

Of course you could make a very good case for including that in the electronics usage totals anyway.

[1] http://large.stanford.edu/courses/2018/ph240/mangu2/

> The biggest users of electricity are things with motors or heaters

I'd guess it's something industrial. Chemicals, in particular, are energy-intensive processes.

Most of the energy in an industrial process would be heat? With second being mixing and other movement?

So that's back to motors and heaters. I cant think of any exceptions apart from electrolysis.

> Most of the energy in an industrial process would be heat?

I don't believe so. When industrial processes require heat, I've usually seen the fuel burned at the plant. Electricity is used for compressors, pumps and such.

You said energy, I assumed electrical energy. So yes most heat probably comes from burning fuel direct.

Would be interesting to know what proportion of electrical energy is used in heating. I would guess its non negligible.

Edit: Making make sense, and addition.

Aluminium production requires lots of electricity. Other metal recycling and heat treating processes too, not to mention extrudation processes.

AFAIK a lot of steel recycling around the world uses electric arc furnaces.

Yeah, electrolysis of aluminum is a huge user of power. You've got the heating to several hundred degrees Celsius and electrolysis combined.

Aluminum smelters. The Hall-Heroult process alone accounts for something like 2% of worldwide electricity consumption.

"Electronics by comparison are a rounding error"

Big rounding error!

By 2040, computers will need more electricity than the world can generate

https://www.theregister.co.uk/2016/07/25/semiconductor_indus...

Double check the table on that article, particularly the scale.

Then as a quick sanity check, consider whether the thing you are actually predicting will in fact use more electricity than the world uses, and still carry on growing exponentially.

Double check before you post if you are really smarter than the guys at ITRS (https://en.wikipedia.org/wiki/International_Technology_Roadm... ) or if you are just maybe incapable to understand the graph.

The scale on the left is orders of magnitude 1.00E+3 is 1000, 1.00E+4 is 10000, do you have a different interpretation?

The benchmark line on the graph shows that by 2040 we'll be using 1.00E22 electricity versus 1.00E21 world usage. So that means we'll be using 10 times more electricity on computing than everything else. And the lines seem to cross at about 2037. Do you think we can increase electricity production 10 fold in 3 years?

We're talking as you say a doubling or tripling of demand for electricity in that case.

The costs to replace the current transmission and distribution grids in the US are estimated at $500 billion and $1800 billion, respectively.

So we're talking ballpark $5 trillion to enable a tripling in electricity demand. Over a ten year period, this corresponds approximately to the entire US defense budget.

We're literally talking about digging up all of the streets, everywhere. I don't want to be negative, but that's what we are talking about.

> We're literally talking about digging up all of the streets, everywhere. I don't want to be negative, but that's what we are talking about.

Aren't the vast majority of power lines above ground?

I was thinking about this very problem the other day and I think it actually may be the best reason why residential storage batteries (as opposed to grid-scale) make sense. It might be easier to keep your transmission system and then trickle charge a whole bunch of small batteries to power end uses.

The increase wouldn't be evenly distributed though. Industrial areas where huge buildings need to be heated and a lot of freight traffic happens might see a much larger increase in power demand than a remote farm.

If we manage to coordinate the charging patterns of cars (for example with electricity prices that change by the minute based on local demand) we might not even need to replace most of the residential grid: Personal vehicles are mostly charged over night when load is currently at 1/4 of the peak handeled in the evening. A lot of other uses like heating can also be shifted with the right incentives to not occur at times of peak usage, further reducing the load increase.

I don’t think that’s actually true, transmission grids are built to handle peak demand, and a doubling or a tripling of total electricity use does not mean a doubling or tripling of peak usage.

In temperate climates for instance charging of cars will almost entirely happen overnight, with a significant rate discount. The basic existence of such a large battery will smooth out the peaks as well as add to them. In fact you’re likely to get far better utilization out of the existing infrastructure, and any new infrastructure which is built, which will bring the per unit costs down.

And in any case, the electricity grid pays for itself, if people are willing to pay the going rate then the infrastructure can be built to satisfy that demand. What is the amount of money spent each year on maintaining and upgrading oil and gas infrastructure? For that matter what is spent maintaining the existing grid? These are huge figures, electricity companies will not be doing this for free, they are happy to maintain the grid in exchange for the revenues that come from it, and they will be only too happy to grow into competitor markets and use what was their competitors’ revenues to fund their expansion.

> transmission grids are built to handle peak demand, and a doubling or a tripling of total electricity use does not mean a doubling or tripling of peak usage.

Well, in theory no. But in practice, if you look at the duck curve [1] of any electrical distribution system, you find that the difference between lowest and highest load during the day is pretty small, maybe 30% of peak load. So really there is a very limited potential to be had by smoothing out the curve, maybe you can get a 20% increase in total electricity use, but certainly not a 200% increase.

[1] https://en.wikipedia.org/wiki/Duck_curve

> What is the amount of money spent each year on maintaining and upgrading oil and gas infrastructure? For that matter what is spent maintaining the existing grid?

Annual maintenance cost for the US T&D grid is $40 billion. We're talking "increase it by a factor of 10". If you assume the power companies pass this cost onto customers with zero profit margin, it's an increase in the electricity price of 9 cents per kWh. Current national average price is 10 cents per kWh. So it's doubling the price of electricity, while saying "we will use 2x-3x as much of it". An increase of 4x to 6x in your electricity bill.

To estimate out how much slack there is in local distribution systems, you want to look at the demand curve rather than the duck curve. Further, you want to look at the peak of this curve on a high-demand day -- for California, that means a hot summer day. For northerly regions it will be a cold winter day.

http://www.caiso.com/TodaysOutlook/Pages/default.aspx

It looks like the minimum CAISO demand today -- during the shoulder season -- was 19 gigawatts.

Here are the CAISO historical peak loads from 1998-2018:

https://www.caiso.com/Documents/CaliforniaISOPeakLoadHistory...

The peak last year was 46 gigawatts on July 25. The peak in 2017 was 50 gigawatts on September 1. So I do think it's fair to say that there is a 2x factor of minimum:maximum use on at least some existing grids.

There's one more issue, of course: how much minimum-period slack is left on the day of high demand conditions? Using the date picker on the demand curve to go back to July 25, 2018, it looks like demand bottomed out at 26 gigawatts around 4:00 AM. There is still a significant diurnal variation in demand that could be used for several hours of nighttime EV charging without increasing system capacity.

The difference between peak load and lowest load change a lot depending on where you look. Geography plays a huge role (heating/cooling running in the night?).

The effect on the individual subgrids can also be much larger than on the grid seen as a whole. For example if you look at [1] (a graph of a hot California day in 1999 taken from [2]) you have commercial peak load at 3 times the low point, residential peak at about 4 times the low point, and industrial peak and agricultural peak at maybe 1.3 times their low points. However because of the way the peaks are offset from each other the total demand only varies by a factor of two. So the spare capacity in off-peak times of any individual power line might be much higher than a graph of total demand suggests.

1: https://www.mpoweruk.com/images/elec_load_demand.gif

2: https://www.mpoweruk.com/electricity_demand.htm

"Current national average price is 10 cents per kWh"

That is amazingly cheap compared other places around the world.

> Annual maintenance cost for the US T&D grid is $40 billion. We're talking "increase it by a factor of 10”

You quoted $2300bn to replace the transmission and distribution system, you say you’d need to increase the $40bn maintenance figure tenfold to replace the grid, but $400bn a year would mean doing that in five and half years. In reality whatever increases turn out to be needed have to happen over thirty to forty years.

And of course doubling demand would be accompanied by huge extra revenues, the electricity system takes in on the order of $400bn a year, doubling the unit production of electricity would mean $12tn of extra revenues, $24tn in total revenues over a 30 year timescale, relative to this it’s difficult to see how $2tn of extra costs on its own can mean a doubling in electricity rates, as you suggest.

That's really not the case, and plenty of utilities have come out saying that residential electric cars won't cause much issue for them. That's not to say there won't be some costs, but it's not nearly so dramatic.

The main issue is peak vs. base load on the grid. BEV charging affects base load, and with variable pricing, won't affect peak load much at all. When you further consider that most infrastructure has some margin to accommodate increased loads, along with on-site generation with rooftop solar and 'powerwall' batteries, you'll see that there's a lot of room for growth.

Even in cases where substantial capacity increases are required, it doesn't amount to a wholesale replacement of current infrastructure. It means adding more 'backbone' transmission capacity, some substations, and a few key distribution overhauls. The expensive 'last mile' stuff won't need significant changes.

Since you seem to actually know what you are talking about, what do you think about nuclear power?

I am pro-nuclear myself but that's mostly because I think it is cool rather than for rational reasons. The problem is that I have a lot of trouble finding reliable data. What I have gathered so far is:

- The capital costs are ridiculously expensive.

- Decomissionning is also ridiculously expensive, and so is maintaining older reactors.

- Fuel is cheap and plentyful.

- The ecological footprint is low, possibly the lowest, except when it goes boom.

- Waste management is an overblown problem.

- It is great for base load (very high capacity factor), not economically viable otherwise.

- Doesn't play well with wind and solar for the previous reason.

Modern plants don't need to be decommissioned. All parts are fully replaceable. Additionally, the point about maintaining older reactors isn't entirely relevant if we are talking about building new ones.

With regards to capital construction costs, there are a number of factors in western countries that seem to drive up costs, and there is a lot of debate around the relative impact of unnecessary litigation that delays project construction, and the enforcement of safety regulations that were designed for older reactors.

There is no arguing which energy source China is investing in most - they have approximately 25 reactors under construction right now.

> Modern plants don't need to be decommissioned. All parts are fully replaceable. Additionally, the point about maintaining older reactors isn't entirely relevant if we are talking about building new ones.

That was my point. You can either extend the life of reactors almost indefinitely, which is very expensive. Or you can build new ones, which is also very expensive.

I heard that new reactors are indeed easier to maintain but to what extent? When every part becomes a radioactive mess, it may be a better idea to properly decommission the plant and build a new one with new technology rather than to change every single part, even if it is possible. There will be a point where it will happen, and it will be expensive, and hopefully, properly budgeted.

Politics certainly play a role in the costs but the plant itself is inherently expensive. For example, the reactor vessel is single 500 tonne part made of a special kind of stainless steel. In fact every part inside the reactor has to be designed to resist radiation. And of course, the safety requirements are huge. I think the EPR has 4 different cooling systems, for redundancy and to avoid having to deal with another Fukushima disaster.

As for China, AFAIK, not only they are investing a lot in nuclear power, but they are very good at it.

Is that an american or global perspective? Specifically the natural gas part. Not all countries have the same resources available.

That is an excellent point, and very difficult to model.

Maybe more importantly, the supply curve for natural gas is also very difficult to model. In Europe NG is expensive, but as the demand drops the price will likely decline sharply (the infrastructure is mostly built, and the current costs are highly skewed towards amortizing investment capital). There are also geopolitical concerns which may play out very differently than expected.

I’m not sure that anyone in Europe is happy about their dependency on Russian NG.

There are other potential sources, most notably, Iran. That comes with its own geopolitical issues, such as US throwing a fit about it. But from a purely European perspective, it's probably better than Russia, since there are fewer conflicting interests.

When articles say energy source X costs Y per kilowatt hour, do those costs include cleanup? For example, does the $0.10/kWh for offshore wind in 2020 include tearing down that nice clean windmill when it wears out?

Grid-scale bids are generally full lifecycle costs. It includes construction, operation, maintenance, and EoL.

Your post doesn't discuss demand response. How would you refute the claims of someone like Jigar Shah who often states that cheap renewables will not need batteries if flexible demand is implemented, examples being heating and cooling electrification and demand flexibility...if the sun and wind are providing extra, use it to super heat or super cool preemptively before the weather shifts.

Because if your approach is to dramatically change the most complicated part of the electrical grid, consumption, and you haven't exhausted the supply side improvements, then you're probably not going to succeed.

I think a lot of that controllable consumption could be done by the computing power of a $6 ESP8266.

It shouldn’t be hard for my fridge or hot water heater to buy a kwh of electricity anytime over the next 7 hours to do it’s thing.

Same for dishwashers and clothes dryers: I’m 100% okay with pressing “Clean/dry within 8 hours” setting.

Same for electric cars: I don’t care if it charges between 2AM and 4 AM or 7AM and 9AM, as long as it’s topped off by 9AM.

I wouldn’t mind a piece of code on my laptop that would switch to battery when the electricity price became massive for a small period of time.

Hot water heaters are effectively thermal banks.

It shouldn’t be hard to convince people to replace their AC with a unit that makes ice overnight if it cuts their electricity bill in half.

If it’s exonomically viable, you’ll have people knocking on your door trying to sell it to you while still being beneficial to the buyer.

I am probably similarly placed - but I am also going to be able to afford to renew things. I'm wondering whether the 99% will be incentivised enough though.

Going forward, it’s simple: Charge the operations costs up-front.

IE: charge up-front for the first 5 years of expected electricity use of the device.

Then it goes against your bill over the next year.

Now everyone buys the cheaper to own+operate one without thinking.

That seems like it would make low income people even less likely to upgrade.

Depends on the design life of the thing. And how expensive it is to operate to begin with. And whether its even a thing that low income people would generally have in the first place.

This is unrealistic:

>It shouldn’t be hard for my fridge or hot water heater to buy a kwh of electricity anytime over the next 7 hours to do it’s thing.

Fridges lose a ton of cool air when you open them. You're going to end up with bad food if refrigeration can't kick in for 7 hours.

>Same for dishwashers and clothes dryers: I’m 100% okay with pressing “Clean/dry within 8 hours” setting.

Waiting 8 hours to dry clothes is a great way to end up with dirty clothes again because of the moisture trapped in from the washer.

>Hot water heaters are effectively thermal banks.

Not really. Unless you effectively turn them into boilers, you max out at boiling temperature. Given that's barely twice the normal setting, it's not going to last for any long period of time that way.

>It shouldn’t be hard to convince people to replace their AC with a unit that makes ice overnight if it cuts their electricity bill in half.

Ice is useless for people that need air conditioning. I think you vastly underestimate how much ice it would require to replace a large air conditioner cooling a single family home by 30F or so for 6+ hours.

For the fridge: you use excess cooling in the freezer to keep the whole thing cool. Or a reservoir of compressed refrigerant. Or phase change materials.

The point wasn’t to be off for 7 hours, but to look at the current and projected prices and decide when to run at a normal rate or when to kick into overdrive (2C fridge and deep freeze) or when to run at normal rate (maintain 4C and -18C).

For the dryer, depending on electricity price, you run it at a slower or faster rate. Indeed, when I line dry, it can take a while, without the issues you bring up.

If there’s a gust of wind, dial it up. When it’s calm, dial it down. Since wind goes in a direction, we have some idea of when turbines will put out more power that’s cheaper.

Hot water heaters here have thermostatic mixing valves, so as long as we’re not boiling, we have a lot of margin to work with without scalding anyone.

Finally, the nice thing about ice is that water is cheap. It’s already a commercialized tech, just not cheap enough for residential use.... yet.

The point isn’t to use these reservoirs for 100% of needs, but size them up to be fully utilized on a regular basis buying electricity at low prices to offset higher prices.

Sometimes that’s less efficient from a thermodynamic standpoint (the hotter hot water heater does shed more heat), but made up for in reduced op costs.

Fridges: energy loss isn't that bad, especially when your fridge is decently filled. Most of the heat capacity is in the contents, not in the air. My current fridge is already off for multiple hours at a time. With a couple degrees of margin and a little bit of high-heat-capacity padding, this is realistic to do.

Clothes dryers: Agree, don't start a wash cycle, then wait 8 hours after finishing to start the drying cycle. But delaying both at once does work.

Hot water heaters: Already being done at a smaller scale, look up Quooker. Given known usage pattern, the replenishment of hot water can also be delayed.

Ice cooling: The math checks out. The temperature difference of 200 gallons of water from 5F to 70F provides over 400.000 BTU of heat capacity. A quick Google tells me that the upper size of air conditioning capacity is 14.000 BTU / hour, so that's more than enough. Given proper isolation, this can probably be lowered by at least an order of magnitude. 20 gallons isn't that much, in my opinion. Besides, this is already being done in practice, using the underlying soil as heat buffer. It has enough capacity for a whole season.

It's not ideal, but mathematically it checks out.

> It shouldn’t be hard for my fridge or hot water heater to buy a kWh of electricity anytime over the next 7 hours to do it’s thing.

I think you'll find the idea of exposing the electricity market directly to consumers is not highly popular with consumers. Not that this is necessarily a bad idea, but definitely not popular right now. If you can think of a good way for the signalling layer of that to work, and find a utility to work with, then certainly it could help, but it isn't going to do everything you want.

> Same for dishwashers and clothes dryers: I’m 100% okay with pressing “Clean/dry within 8 hours” setting.

Maybe you don't do laundry that often, but leaving wet clothing in a dryer for eight hours is not a good idea, in my experience.

> It shouldn’t be hard to convince people to replace their AC with a unit that makes ice overnight if it cuts their electricity bill in half.

Due to thermodynamics, this is not helpful.

> Same for electric cars: I don’t care if it charges between 2AM and 4 AM or 7AM and 9AM, as long as it’s topped off by 9AM.

That's mostly new load (as far as the electrical utilities are concerned), so it isn't helping reduce peak demand.

Flexible demand doesn't go far enough. There are very cold weeks with little wind or sun (the sky might be clear, but because of angles there isn't much sun anyway). We would have to heat our houses to over 100C to keep the pipes from freezing before the end of the cold spell. Similar problems exist in summer.

Demand can help a little in that we can heat houses warming in the evening while the wind is blowing and let the house drift down all night.

Of course better insulation is assumed, but then you need an air exchanger which needs power to run as well.

You claim the author is cherry-picking, and then you say this:

> The marginal cost for natural gas is roughly ~1.8 cents/kWh today (including O&M)

I am assuming you worked this out using the Henry Hub prices of circa $2.50/mmscf - yes? But hardly any gas consumers pay these prices, even in the US. In Florida and New York, for example, the prices are more like $5.00/mmscf, and can be even higher. And in the rest of the world - which accounts for the vast majority of electricity production - the prices are far higher still, commonly around $10/mmscf, today.

>solar would have to be ~$0.50/W ... ... that is ~5 doublings from today

Not really sure what $0.50/W means here - is it $0.50 CAPEX per "peak watt"? If so: The cost of utility scale solar power plants today is around $1.00/peak watt, so with your assumption of a 20% learning curve it will only take 3 doublings to reach that.

do you publish reports / models / analysis ? I've been eager to find such documents for years

How do you do industrial heating without natural gas? Last I heard heating with electricity was ridiculously expensive.

High temperature nuclear reactors?

More likely process heat from SMRs and other compact nuclear reactors.

Hydro is an option, albeit limited by location

Define industrial heating?

Low grade heat can be got from heat pumps. It can be topped up to a higher temperature with resistive heating.

You can also lower demand for heat with heat exchangers, insulation and such.

So a chemical process that required 50c would be fairly easy to do. Smelting metals gets harder.

That's an interesting topic. I wonder if the cooling down of steel in a steel mill can be used to increase the temperature of the smelting pots.

For an integrated Steel mill most of the heat comes from chemical energy (i.e exothermic reactions).

In the blast furnace this comes from Coke, in BOS/LD Converter alloys such as ferrosilicon or silicon carbide "heat raisers" can be used to increase the temperature.

Modern plants have things like cogeneration and recovery turbines to produce electricity from the off gas.

Biggest electricity usage tends to be from things like extraction fans and conveyor belt motors.

If we're going to drastically reduce CO2 emissions, we'll probably have to replace blast furnaces with something like hydrogen direct reduction, which needs boatloads of electricity to produce the hydrogen.

There's an aluminium smelter a 100kms or so away from where I live which is using electricity for smelting. It does use a lot of electricity but by being such a large consumer, it gets very large discounts.

But for this same reason, we have deliberately built them in places where electricity is cheap, often right next to hydro power plants. I don't think existing factories that are converted could get deals that are quite as good.

Industrial heating is one of the solutions to the problem, because a lot of processes are not that time-critical. Some processes already require electricity, such as arc furnaces. However, as the process cost is dominated by the electricity usage, they are already being run at times when the electricity cost is very low, that is, when there is overcapacity. This makes it easy to absorb overproduction of a grid which is intentionally oversized for peak demand. It's already not unheard of for industries to get paid to consume electricity.

Definitely nuclear. But you need to have a plant size that is right. That is, you shouldn't have grid scale reactors for this, as the power source has to be within a couple of miles of the factory/process location. So the size of the reactor has to be on the order of 10 to 50 mW. And you need a good method for getting the heat there - probably a molten salt intermediate heat transfer system... There are a couple of nuclear companies doing this seriously.

You are probably looking at wholesale electricity production. At the retail end of the line things look different. But first a disclaimer: I'm an Australian, and we pay around AUD$0.27/kW hr. It varies a bit - one state it's closer to AU$0.36/kW hr.

Right now if you buy a PV + battery system that will cover all your usage, you are looking at $20K before rebates [0]. If that drops to $10K it becomes better for households to switch from being users to generators - where "better" means better than investing the cash and getting a 5% return after inflation over 10 years. In other words it's a no brainier. If a household doesn't choose to do it themselves, someone will come knocking on their door and pay them to do it.

For Australia that is the tipping point. Once we cross it the electricity market will change forever into something unrecognisable. There are 9M houses in Australia which when fitted with today "standard" 6.6 kW system with 5 kW inverter will generate 50% of our daily electricity usage. But it won't just be profitable for houses of course - it's not like businesses don't have roof space too. It's not too hard to see that other 50% disappearing too.

The thing preventing it from happening right now is battery prices. Battery are ignored as far too expensive by people looking at grid storage and right now they are too expensive for households too - but if that 8% per year improvement figure being bandied about is correct we will hit that $10k threshold in 10 years.

It won't suddenly happen in 10 years. We've are seeing close to EOL coal fired generators bought by purchasers saying they would squeeze another 20 or 30 years out them later shut down not long after. I'm not sure why, but the pattern has been repeated several times now. I do know in Australia the wholesale electricity price used to go negative most nights. Now we don't know if that still happens because our conservative government shut down the reporting of the wholesale price, however I suspect it's worse - not only does it go negative at night, it must also go negative at when the wind blows hard. Having to pay others to dig the coal out of the ground because you can't adjust quickly enough must be painful.

Regardless of why they're shut down, it's now pretty obvious every time a coal fired plant shuts down the prices go up the next year because it's happened several times. And so more people buy solar - we have the highest household solar penetration in the world. While the price goes does up less people pay it, overall less money comes in and around and around we go. The pace seems to be accelerating, and it will continue to accelerate until the wholesale price starts dropping. When most of your existing generation comes from coal, and coal generators are being retired every couple of years, it's hard to see that happening any time soon.

I don't know why the focus is always on the prices wholesale end of the chain. The big change isn't that renewables are slowly catching up to wholesale prices - the big change is they have already passed retail prices. While putting a coal fired generator or wind turbine in your back year was never feasible, throwing a few solar panels on your roof and installing a battery is feasible. Sure, the power generated costs 2x what a wholesaler can generate it for - but in Australia the price the wholesaler gets jumps by a factor of 3x to 4x. If battery prices keep dropping, there will be enough space for them to fit right in.

[0] It's actually used $20K for this type of system right now. After rebates it's about $13K right now, but when households are threatening to wipe out 50% of the generation market I expect the rebates will disappear.

[1] Average wholesale price: AUD$0.073 https://www.aemc.gov.au/energy-system/electricity/electricit...

[2] Average retail price: AUD$0.27 https://www.canstarblue.com.au/electricity/electricity-costs...