The future of electricity is local

I frequently read about proposals for new solar power developments where the resulting power is moved great distances to less sunny places, such as northern Europe from the Sahara, the US North East from the South West, or even Australia to Singapore. According to these proposals, the key to unlocking greener electricity is massive infrastructure to move renewable electricity across continents.

Looking at a solar resource map, such as this one of the US, it certainly seems that if the solar resource was oil, Arizona would be the place to drill.

On a global scale, it seems that solar potential is reasonably localized, with as much as 3x more sun in North Africa compared to Scotland. Furthermore, seasonal variations are more extreme at high latitudes, accentuating the relative advantages of long-distance power transmission.

And yet I believe this discussion misses an important detail. When we talk about the future of electricity generation we’re talking almost entirely about wind, solar, and battery storage. Biofuels rely on high impact farming and low efficiency photosynthesis, while hydroelectric power is strongly dependent on available water and geographic conditions. And despite mostly fake political disputes, coal is very, very dead. Like solar, wind is also localized in terms of optimum placement, as some parts of the world are reliably very windy at the right times of day for human power demand.

But most importantly, these three technologies are a) already cheaper than any existing method of power generation, even for new builds and b) continuing to get cheaper at a frantic pace.

https://arstechnica.com/science/2019/08/wind-power-prices-now-lower-than-the-cost-of-natural-gas/

In areas with both good wind and sun, wind is still incrementally cheaper than solar but solar’s rate of improvement is faster. In the long run, I believe wind will only be deployed in solar’s gaps: places where it’s reliably windy after dark (such as the great plains east of the Rocky Mountains), or it’s dark even during the day (such as the North Sea).

In terms of land use, I’ve written before (and I’m not the only one) that wind and solar are more lucrative and generate more power than oil drilling and coal mining respectively. Even if the UK found another huge oil field in the North Sea, building wind turbines would generate a bigger, and faster, rate of return. Even if Australia found a new 3 m seam of coal in the Hunter region, just beneath the surface, using it as a foundation for a solar plant would generate a bigger and faster rate of return than digging it up and burning it.

So far, so good, and with ongoing 5-10%/year cost improvements, countries the world over are prematurely decommissioning coal fired power plants built directly adjacent to productive coal seams to build solar. The Paris agreement doesn’t look so unachievable in this light.

There is still a missing piece of the puzzle, and that is intermittency. With modern weather prediction, both solar and wind production is substantially more reliable than coal or gas generation, but there still are times of day when demand outstrips supply, particularly in the early evening. There has been unending wailing and gnashing of teeth about this problem, particularly because nearly all forms of grid-scale energy storage have been impossibly expensive since Edison and Tesla built the first grids.

This is not the case anymore. Tesla’s big battery in South Australia made a lot more money than anyone seemed to anticipate, paying for itself in just the first 30 months of operation. Usually, if a big new utility infrastructure project is unexpectedly profitable, that’s a HUGE surprise. Utility megaprojects usually lose money, sometimes surprisingly slowly.

As I have written before, it turns out that in the frequency control auxiliary services (FCAS) market the provider with the fastest response can completely control the market, forcing other providers to transfer almost unlimited wealth in any desired direction. Batteries are also being routinely deployed in, among other places, Los Angeles for load shifting. In addition to being able to provide FCAS services, it turns out that utility scale lithium-ion batteries are a financially viable way of saving up power during the day and using it in the evening. Local company 8 Minute Energy has won several bids to build huge solar+battery projects in the Mojave at a combined price of $40/MWh, more than five times cheaper than the next best gas-powered solution.

For a long time, it seemed that of wind, solar, and batteries, batteries would be the biggest drag of the three. And yet two years ago CleanTechnica reported on stunning levels of investment in battery production, with projected battery production climbing 10% per month. More recent reports show that battery costs have dropped by 88% in just ten years. And just like the other two, with long term investments both growing and compounding, these cost reductions show no signs of slowing down. Nor is there any physical reason to expect that cost reductions will plateau in the foreseeable future.

The future of electricity generation seems fairly clear. Over the next decade I think it’s safe to say that existing investments lock in 5% annual cost improvement in wind, 10% in solar, and 15-20% in batteries.

How does high voltage transmission figure into all of this? The future of power prices is more exciting than the past. During a sunny day, solar production may exceed demand, leading to zero (or even negative) prices. In the evening, batteries provide power at a certain premium. If batteries are exhausted then more expensive gas plants need to spool up, and their costs are increased by less usage due to amortization.

Jumping into these opportunities for real time arbitrage is the possibility to import electricity from other places a long way away. New York’s evening power supply can be met by late afternoon solar in New Mexico or an evening breeze in northern Texas. At the point of generation, the power is as cheap as solar is anywhere the sun shines, which is to say ludicrously cheap. At the point of use, there is unmet demand and prices may be as high as $200/MWh, which is the cost of running gas generation. If long distance transmission can deliver power more cheaply than gas, the game is on! Can it?

This source suggests a capex cost of about $1000/MW.km for long distance transmission lines, with some variations for a bunch of reasons. The distance from Arizona to New York City is about 3000 km, which is comparable to other long distance power transmission cases. Therefore the fixed cost of power works out to be about $3m/MW. At $200/MWh, the transmission line would have to sell 15,000 hours of 100% capacity to reach breakeven. Factoring in capital costs would double that. If New York needed the equivalent of 2 hours of power imports for 200 days of the year, then it would take 75 years of operation to bring in that much power. 400 hours a year (compared to 8760 hours in a year) is less than a 5% duty cycle, which is clearly inadequate to deliver any kind of return.

Perhaps the line could be used for baseload supply during more ordinary times? Recall that 8 Minute Energy is selling projects in LA for $40/MWh. If the solar resource in New York is half that of Arizona, then perhaps $80/MWh is a realistic cost for local solar power with similar load shifting capacity. Can transmission turn a profit with a $40/MWh cost difference? With constant 100% usage, capex and capital costs of $6m/MW would be met with just over 17 years of operation. Of course, 100% usage doesn’t reflect a factor 2-4 diurnal variation in consumer electricity demand but this is an intentionally optimistic model. Perhaps 17 years is enough to make the business case close? Particularly if your company makes money during the construction phase?

Ah, but we are missing an important detail. That $40/MWh cost difference is a maximum. If people are making money selling electricity to New Yorkers for $80/MWh, then imagine the cost pressure if people could do it for $70/MWh. And yet at present rates of cost improvement, that’s less than 18 months away.

Indeed, electricity in New York has long been a bit more expensive than the national average, at around $200/MWh for the consumer. Let’s say that the solar resource is relatively poor and comparatively seasonal, making it 5x worse than Arizona, where electricity is about $128/MWh for the consumer. Combining these two effects gives a factor of 3.2 in favor of Arizona. That is, solar+batteries have to be 3.2 times cheaper in New York to have comparable economics in Arizona.

Solar power is getting 10% cheaper a year. At that rate, a factor of 3.2 is equivalent to waiting just 12 years. Solar has been overwhelmingly competitive in Arizona since about 2013 (despite the relative infancy of batteries back then), so by 2025 we might expect to see solar deployments begin to pick up in the North East.

This might seem crazy that places as cold and dreary as Boston could employ solar power simply by building 5x or 10x nameplate capacity, but it turns out that solar is so cheap that overbuilding is the cheapest option, and project economics strongly favor solar over any other kind of power plant or transmission line, all of which have rates of return measured in decades rather than years.

Let’s return to the transmission business case. In this case, some fixed infrastructure with front-loaded costs is trying to eke out a profit in a precipitously falling market. Recall that at $40/MWh margin 100% capacity would see a profit in just 17 years. What happens if that $40/MWh gets squeezed by 10% per year? Now, after 17 years, the power line has made just $3m/MW, less than half the previous estimate. Indeed, after that revenue disappears almost entirely, with lifetime revenue asymptoting to $3.5m/MW. And this is assuming preferential access to the market and 100% utilization. It’s dead in the water before the first foundation is poured.

While I’ve focused on a particular use case here, in the US, the observation generalizes. Any given fixed infrastructure will have a cost floor below which utilization will not generate any value, and steady advances in wind, solar, and batteries will eventually break through that floor. For local transmission in particularly gnarly places, power transmission may be profitable well into the 2030s. But in general, as the costs of local generation and storage continue to plummet, the relative inefficiency of long distance transmission will drive its use out of favor. This will mean the proliferation of rooftop solar and consumer batteries, as well as solar+batteries at utility scale, as well as municipal batteries in containerized form factors wherever local energy value arbitrage can justify their use.

I think it’s safe to predict, therefore, that the future of electricity will see mostly local generation and consumption, with multiple federated smaller grids and distributed energy resources (DERs) forming a flexible, syncretic, and robust system. I’m all for major public investments in infrastructure, but let’s spend it wisely.