Brian Bartholomew tweets a lot about grid decarbonization. A few days ago, he tweeted this:
Brian is correct in pointing out this rather egregious structural flaw in the United States’ grid infrastructure. Essentially, because there are three distinct “power grids” (the East, West, & Texas) within the country with very little physical interconnection, power from one cannot flow into another. The imbalance in supply (and therefore price) that can result from this balkanization can be jarring.
It can also be quite dangerous – in 2021, Texas’ power grid became extremely compromised during a historic freeze. All sources of power within the state suffered a decrease as a result of the weather, particularly the state’s primary method of power production, natural gas. A power grid must run roughly at the same level of supply and demand because otherwise the frequency of the alternating current running through it could change significantly and damage equipment. Essentially, the whole grid is a delicately balanced rotating machine, somewhat like the pistons in a car engine. During the freeze, grid operator ERCOT (Electric Reliability Council of Texas) throttled the amount of demand that the grid would see in an effort to prevent it from being permanently damaged [1]. While taking power away from people in the freeze is nonsense, the reality is that the Texas grid can only pull a few hundred MW from adjacent power grids, a small percentage of the ~60,000 MW peak demand the grid usually sees.
The grid situation in the US also has had significant implications for the effort to transition away from fossil fuels, such as coal and natural gas turbine power plants, to renewable sources, such as wind and solar. When moving from the former power source to the latter, the energy generated in a given area is subject to the current conditions of the local environment. If it’s cloudy, you will get less power from solar panels. This becomes a headache when you need to match supply to power demand and it is known as the intermittency problem. Perhaps it is cloudy in California and windy in Texas, as it often is. Without robust connections between different power grids that could “smooth out” this difference in local supply and demand, the only solution is often to fire up the reliable natural gas turbines to make up the difference at a moment’s notice. The more power generation and supply connected to a single grid the less fragile that grid is – but cost, political, and technical hurdles lead to a tradeoff to be considered.
In some of the ensuing discussion around Brian’s original tweet, Casey Handmer tweeted this in response to the idea that a nation-scale power grid is needed:
Casey’s point is essentially that building the power lines required to connect the grid at national scale (high-voltage lines of DC or AC) is much too costly at the energy arbitrage that it would allow for it to make sense relative to the cost of local storage (battery) projects. He claims the tipping point for this is about 140 miles, which is to imply that any power transmission in the U.S. over around this distance is uneconomical. We build power lines of this distance often, so I wanted to prove his math for myself.
If you assume High-Voltage DC (HVDC) lines would be built to transmit power across the states, you end up with around $1M/MW (or ~$850/MW-km). Notably, it seems that the power capacity of these high-voltage lines makes them fairly cost-effective over long distances. If you look at the capital costs of some more standard lines (around 345 kV), these lines are usually around 200 km or so and max out around 700 MW. This gives about $2000/MW-km [2]. High Voltage DC lines are cheaper because they require fewer conductors, less metal for towers, and lower land costs. Of course, the reason we don’t build HVDC lines for everything is because the equipment they require is more expensive, but this just gets amortized over the longer distances that these lines cover. Another assumption Casey makes here is that the battery capital cost is somewhere around $200,000/MWh, which looks to be fairly conservative in light of recent surveys on battery costs, but fairly optimistic according to power regulators. That being said, this price has more than halved in the past decade, which is why Casey claims the critical distance of 140 miles is falling by 10% each year.
Below is a sensitivity analysis of the critical distance (in km to annoy people who use miles):
The tipping point from the tweet is boxed
Interestingly, the line connection cost on the upper end, where typical transmission line cost lives, the critical distance is even lower! This is implying that power lines longer than ~100km are not economical to build given a battery alternative. This is another indicator that something is slightly off with this model.
Beyond the solid battery and power line cost assumptions, there are two more factors that underlie this model: the power arbitrage (or how much $ you make for selling the power you bought in a different market) and the number of cycles (or amount of time that you’ll be selling this power). What this model misses is that the cost of the power of a battery is the local marginal price, so the battery is not necessarily subject to the arbitrage that high-voltage lines would be. For example, if I live somewhere where power is very expensive, I’m likely going to have to charge my battery for roughly the cost of that expensive power. What Casey is assuming here is that the batteries can be charged up some amount of hours prior to discharge for $100s/MWh cheaper than that power will be sold at. For an area where lots of power is needed for days on end and those batteries cannot be charged inexpensively, this critical transmission distance effectively increases for a proportional decrease in the cost of the power stored in batteries.
Of course, this can be mitigated by bigger, cheaper, and more batteries but there is a practical upper limit, just as there is a practical upper limit to transmission distance. Say you needed to store power for 10 cycles (5 days) instead of half a day. You’d need up to 10x the battery MWh, which would effectively 10x the critical transmission distance allowed.
Transmission of Power Over Long Distances
The idea of a critical distance also applies to transmission of power via AC or DC. Over long distances, the main problem is the voltage drop that can occur from losses. For traditional distances (less than ~600km), AC is best because higher voltage can be run to avoid transmission losses over a distance and transformers can be used to convert to lower voltages for use at an end point. DC isn’t subject to the capacitive or radiation losses that AC is, but the cost to convert DC to AC for use at the endpoint is tremendous so no one does that for short distances.
HVDC Xilin Inverter Plant, China – this is what it takes to go from DC to AC
Once the threshold is crossed into longer distances, it becomes worth it to build these conversion stations for the lower transmission losses at incredibly high voltages.
In particular, China builds a lot of these HVDC transmission lines. The bulk of it’s population lives in the east, physically quite far from the hydropower in the west and coal in the northwest. Therefore, ultra-high voltage (UHV) transmission has been key in a country aggressively stepping up its power consumption. In the past 20 years, China has grown from an installed generation capacity of 443 GW to about 2,000 GW, more than a 3rd of which is renewable. For reference, the installed capacity of the United States is around 1,100 GW. In the same amount of time, China has built 35 UHV lines spanning ~35000 km, transmitting about 450 TWh with a total capacity around 50 GW and the country plans to double this capacity by 2025. This is done relatively inexpensively as well. The record breaking Changji-Guquan UHVDC Link, connecting western generation to eastern population centers 3,324 km at 1,100 kV. It cost about $5bn, so ~$200/MW-km – that’s about a fourth of the cost I assumed for the United States.
The United States struggles to build simple low-voltage transmission lines, much less UHV lines. The most notable recent UHV project in the US, the Plains & Eastern Clean Line, is a 4 GW line from Tennessee to Oklahoma that was held up in development for 8 years before being effectively cancelled in 2017. This is very bad, not only because UHV lines could allow the US to reduce it’s CO2 emissions by 80%, but also because renewable power is on track to become extremely cheap. Without the UHV interconnects, the US runs the risk of not being able to economically build out renewable power and being left behind. Biden’s IRA allocates around $3bn to the buildout of more power lines, specifically for linking renewable generation to loads, but this must be merely a start.
Batteries Are Winning
The whole point of the UHV interconnects is to “smooth” out the power generation and power load mismatch by spreading it over a geographically larger area. Batteries are useful alternatives as storage devices and as mentioned, they get more useful the more of them you have. One interesting way the US market might be sidestepping the intense capital costs of UHV transmissions lines while still pursuing a renewables transition full force is the Vehicle-to-Grid (V2G) concept.
Electric vehicles require about 20 kWh per 100 km (Tesla Model 3 is around 12 kWh/100 km), which is about 66% of the average daily power consumption of an American household (~30 kWh/day). Barring concerns about the insane increase in the load that 100% electric vehicle adoption (we are currently <10% in the US), this amount of battery capacity is potentially a very good thing. Electric vehicles are a massive fleet of distributed electricity storage capacity. If these car batteries can be charged by the grid, they can also be made to give power back to the grid with bi-directional charging hardware. This V2G would flatten the peak fluctuations in grid power demand immensely, which means less dramatic swings in energy prices, and any increase in storage capacity increases the grid’s capacity to utilize renewable energy sources. Many companies, such as Ford, offer this as a standard $1k accessory now. Consumers are incentivized here because energy can be bought when it is cheap and used later when it is expensive. If the grid’s problem is simply that more batteries are needed, Americans are solving that just by buying new cars, which Americans love to do.
Wrapping Up
The earlier suggestion that power lines longer than ~250 km are less economical to build than increased battery storage capacity hinged on the assumption that battery storage could be arbitraged at the same rate as HVDC power. Because the power demand can fluctuate locally on scales of several charging cycles for a single battery, this arbitrage assumption is predicated on a critical amount of battery storage capacity already existing. This is a flywheel, where the more batteries built, the more cost effective it is to build more batteries.
While the US has done a poor job in the past 2-3 decades keeping up with power transmission infrastructure, the rapid adoption of electric vehicles at scale and leveraging these products as distributed electricity storage systems have immense promise. V2G could serve to leapfrog the power grid beyond the capacity threshold where it simply does not make economic sense to spend on massive ultra-high voltage transmission systems, much less any power line longer than a few hundred kilometers.
Augmentation
[1] I couldn’t find any examples of damaged power generation equipment, but here is a LinkedIn post (forgive me) that goes into detail about how this works.
[2] I assumed a 345 kV line costs about $2.5mm/mi
To Be Determined 
Leave a comment