Match Game

Monday 15 May 2023

The time is approaching when the United States will want to ensure that new demand coming onto the grid is matched by growth in clean generation. Currently, for example, no one has suggested such a tightly managed scheme for the ascent of electric vehicles; and one reason may be that the US is creating so much new clean generation each year, that these additions not only match, but dwarf the early ascent of power demand arriving from new, on-road EV. But EV are not the only devices that will soon be coming to the grid. And already, there is evidence that the long plateau of US power demand is breaking out to the upside. The chart below is handy therefore in two ways: total generation broke out to a new all time high last year above 4300 TWh, but thankfully, marginal growth in supply is largely being met by wind and solar. So far, so good. (clicking on chart takes you to an interactive version).

The challenge: it’s not clear how much longer the US can simply deploy new clean generation, and hope that it exceeds new demand coming onto the grid. While 4 TWh are required for every million EV that hit the road, so far, that’s been effortless to match with new wind and solar. And after all, we are running a 4300 TWh power system. But the the dictum to electrify everything means that heating and industrial processes will also migrate to the grid. And one of those new sources of demand will be energy intensive hydrogen production.

The US produces roughly 10 million metric tons of hydrogen each year, but does so using mostly dirty sources of energy, mainly natural gas. Were future hydrogen production to branch more quickly towards the process of electrolysis, however, the electricity sourced to run that process would mostly reflect the grid mix at the time, not so much nationally, but locally. To draw once again upon the example of EV adoption, we know from Argonne National Lab data that an EV running one mile in Los Angeles is cleaner, systemically, than an EV running one mile in Louisville, Kentucky because California’s grid is meaningfully cleaner than in the Ohio river valley region. Accordingly, if you deployed an electrolyzer to make hydrogen in Pasadena, CA, the hydrogen it produces would be cleaner than its counterpart running in Paducah, KY.

But not that much cleaner. And, still not clean enough to call that hydrogen green. This difference matters now especially, because the US government has created a number of initiatives to catalyze hydrogen development, with a notable subsidy to incentivize green hydrogen through a production tax credit of $3.00 per kilogram. That subsidy is quite substantial. And it easily makes the higher cost of producing green hydrogen competitive with current production methods.

But to earn that subsidy, there should be rules, don’t you think? Why should you get that full subsidy for hydrogen production using the current grid mix? Obviously, you shouldn’t. But what if you pair an electrolyzer to a clean source, like nuclear? Surely that should qualify, right? Well, maybe. This very question arose when Constellation Energy earlier this year paired a hydrolyzer with its Nine Mile Point Nuclear Plant in Oswego, New York. Power generated from nuclear is of course 100% clean, and better still, nuclear power plants demonstrate high capacity factors, running nearly all the hours of a year (well, ok, 92% of those hours). The problem is that as Constellation’s hydrogen production runs, drawing clean power from the nuclear plant, elsewhere in the region demand must be met through an increase, however slight, from other sources of power—mostly natural gas. Perhaps you think this is a minor quibble. And, with green hydrogen production in its earliest phase of development, who could care—at least in the near term?

Well, these questions are coming to a head right now because industry wants to move aggressively to take advantage of the green hydrogen credit. Next Era Energy, for example, announced recently it wants to make a colossal bet on hydrogen. But Next Era is also not pleased with the idea that it’s necessary to fully match hydrogen production with clean power on an hour to hour basis, and would prefer an annual target, or measure. In one sense, this sounds reasonable. But what happens when myriad companies rush to produce hydrogen, and grid demand really starts to scale?

This is where academic analysis is useful, and Professors Leah Stokes of UC Santa Barbara and Jesse Jenkins of Princeton have recently weighed in on the matter. In a New York Times op-ed, Stokes lays out the three basic principles that should govern eligibility for the production credit. First, hydrogen should be produced from newly deployed clean power; second, this power should be produced locally; and finally, production should occur during those times of the day clean power is on the grid. Stokes also points out that hydrogen production (like other flexible demand) can pair nicely to those times when solar is in local surplus during the day, or wind power is in surplus, typically at night. These principles are highly relevant at the moment, because the US Treasury Department is about to perfect the rules around green hydrogen production, and as Jesse Jenkins of Princeton points out in this twitter thread, it’s crucial to get it right.

What we’ve learned from history is that strong policy subsidies can cause new technologies to scale quite rapidly, from zero-to-one, if you will. This was true for utility scale solar and wind power, and despite the fact that hydrogen is both early in its development (even earlier in terms of its distribution and use) we should assume that the hydrogen PTC (production tax credit) will land with force. While the current US output of hydrogen through direct, dirty sources will take time to transition to a cleaner method, quantifying current production in electricity terms can give us a starting point to understand the scale of the challenge.

Producing a kilogram of hydrogen through electrolysis requires about 50-55 kWh of electricity. If the US is producing 10 million metric tons of hydrogen each year, that is the equivalent, in term of kilograms, to 10 billion kg (a metric tons contains 1000 kg). And given that each of those kg requires 50-55 kWh of power input, that would translate to 500-550 TWh of electricity each year. That is huge. Again, consider the energy intensity here. A million EV can drive for a year and roughly demand just 4 TWh of electricity, using the standard 4,000 kWh of annual demand for each one of those vehicles. Producing just one hundred kilograms of green hydrogen in a year would, by comparison, require 5000-5500 kWh. 100 kg of hydrogen is not alot of hydrogen at all, in industrial terms. But look at the energy requirements to produce that small volume of hydrogen.

Now there’s a concurrent argument emerging that creating immaculate green hydrogen from the start is too onerous because hydrogen is in its infancy, as an energy source that can replaces gas and coal in industrial processes. Let’s pick that apart. Hydrogen has been called the swiss-army knife of energy sources, but that framing is less helpful than it appears. In one sense, it promises too much. In another sense, it does indicate the myriad opportunities for hydrogen. But hydrogen lacks a global distribution network, and that makes matching production with demand difficult. Europe, for example, already plans to build hydrogen pipelines for this very reason.

But what is not difficult, or in its infancy, is actual production. Requirements that green hydrogen production meet those three tests cited previously are clearly the right idea. Consider: dirty hydrogen is already in production, the tax credit is extremely generous, and we are still able to build lots of new wind and solar. For example, in 2022, the US generated 639 TWh of electricity just from wind and solar, but 98 of those TWh were created in 2022 alone. New EV hit the road nationally and bumped up grid demand by roughly 3.7 TWh. And green hydrogen production is still so small that it would have hardly registered. The US is producing plenty of new clean generation to more than cover new demand, not only from new EV, but from new green hydrogen. For now.

Some will not like this suggestion, but perhaps a compromise that lays out a phase-in period for green hydrogen production would be appropriate. This would allow for an initial period of production in which some producers, admittedly working on a dirt(ier) grid, could purchase renewable energy credits from other regions to come up to the subsidy requirement. Such a scheme would essentially imply that while the US is creating gobs of new clean electricity from wind and solar each year, this new generation is not always in the right place where hydrogen production can more easily meet demand. (Here, think of demand as a kind of demand center, a physical location where an industrial user can take all the hydrogen output, and more.)

When we think about how wind and solar is currently distributed in the US, for example, this problem of location becomes more clear. The Great Plains states are so rich in wind power, and so sparsely populated, that states like Iowa, Kansas, and the Dakotas would be natural places to set up production. But then, how much demand for hydrogen exists in the great plains?

A review of how the US uses its 4300 annual TWh of electricity also reveals that a quarter of this output is used by just four states: Texas (436 TWh), California (247 TWh), Florida (242 TWh) and Ohio (148 TWh). Building more clean generation in two of those states, California and Texas, is still relatively easy. Indeed, as readers know, Texas after ramping itself to the global level of wind power output is now about to pursue solar deployment aggressively. But Florida and Ohio, it must be said, are not as easy.

The US Department of Energy recognizes that pairing production centers with demand centers for hydrogen is a key part of the hydrogen effort, and advanced that challenge prominently two years ago in its kickoff to the Hydrogen Earthshot. There are two main risks therefore to how we govern these early days of hydrogen production in the US. First, you can’t let green hydrogen production take off too far at scale without guardrails and rules, because then you create a legacy set of production infrastructure that depends, in part, on rule relaxation. Worse, hydrogen production is clearly quite energy intensive. So scaling would be felt quickly in the power system. And that would not be so healthy. Second, if the rules are too stringent from the get-go, than the shiny prize of the production tax credit probably spurs something less than a big bang of hydrogen production. But perhaps going slower, ramping up over a longer timeline, is optimal.

These are of course not the only two choices before us, but they do describe the problem thematically. And deciding will not be easy.

Editor’s note 1: The Gregor Letter realizes it has used these particular charts here recently, despite a preference to avoid, in general, such repetition. Sometimes however a chart is worth not just a thousand words, but two thousand words.

Editor’s note 2: Feel free to inspect my math around current US hydrogen production, and the power needs to convert it to electricity at Google Sheets.

Further reading:

Hydrogen’s Power Grid Demands Under Scrutiny in Tax Credit, from Bloomberg Law.