50 years of green hydrogen failure

 From Michael Liebreich.

It is over fifty years since Japan's "Sunshine Project" started governments pouring money into green hydrogen as a replacement for fossil fuels. How is it going? TLDR: not well.  That’s why it’s called the #HydrogenSoufflé.

I don’t even need to write a commentary - I’ll just leave you with five charts:

• We’re not producing green hydrogen (despite the investment
• We’re not buying hydrogen vehicles (despite the hype)
• We’re not buying hydrogen fork lift trucks (despite the rumours)
• We’re not buying hydrogen trucks (despite the hopes)
• We’re not buying hydrogen boilers (despite the grift)

Hydrogen is hard to store, because its small molecules can escape through metal and plastic molecules of its container.  It makes pipes brittle.  It needs to be chilled and compressed for storage, so there are energy costs/losses.  And in most cases, alternative technologies are cheaper and more effective.  For example, the round-trip efficiency of lithium-ion batteries for storage is much higher than electrolysing water and then using the hydrogen in a fuel cell.  For aircraft, the hydrogen has to be kept at  -253 C as a liquid or at 5,000 to 10,000 pounds/per square inch (350 - 700 bar, 1 bar equals one atmosphere).  Tricky.  It can be used to smelt steel, but that works best if the hydrolyser is close to the steel mill.  And as I've mentioned often before, if it is converted to methane by the Sabatier process, it can be used for long-duration storage for balancing the grid.

In my opinion, and obviously also Liebreich's, green hydrogen is mostly hype.  Green methane, maybe.  And as the first chart shows,  only 0.1% of hydrogen produced is in fact green.

Why Trump and big oil won't win

 This is my chart using data from Our World In Data of the price of PV panels in US$ per watt, in constant 2024 dollars.  In other words, a 5000 kW system, ignoring inverter, grid connections, land, and installation would cost 5000x30 cents, or $1500.  (Of course, this pricing is for wholesale systems with economies of scale; a rooftop solar system would be more expensive.)   This is a 99.8% fall from 1975.

Note logarithmic scale

On its own, this is not enough to show that solar will provide most of our power, inevitably, eventually.  After all, the denialists will gleefully tell you the sun doesn't shine at night (goodness me, who knew?) 

So you've got to add the cost of storage.  And the fact is, battery prices are falling even faster than PV prices. 

This chart shows BNEF's battery costs survey data, also in constant dollars, with my estimates for 2025 and 2026.  If you do the numbers, it turns out that adding 4 hours of storage to a solar farm will add just $12/MWh of electricity generated to the cost.  Adding 8 hours storage would cost $24/MWh, which is still cheaper than new coal, or (outside the US) new gas.

Note log scale

But, I hear the denialists wail, what about dunkeflaute, those periods in high latitudes when there is no wind, and little solar, and it's cold?  Well, until we get long-term storage, we will need gas peaking.  We can make the gas using electrolysis of water, and using the hydrogen produced to make methane via the Sabatier system, which would in effect be long-term storage.  Or we can go on using fossil gas.  But even if we do the latter, we will still have cut emissions from electricity generation by 95%.  

A final chart from Our World in Data.  It shows a classic "learning curve".  A new technology starts.  It's expensive, and has only a few uses out in the wild.   But usage increases.  Manufacturers get a bit better at making it.  Demand increases, costs fall.  Falling costs lead to still more demand, which in turn leads to still lower prices, and so on, until the technology has gained a 100% market share. 

The chart uses a double log scale.  On the vertical axis, each tick mark shows a halving of PV module prices.   On the horizontal axis, each tick mark shows a 10-fold increase in cumulative installations.  So each 10-fold increase in installations leads to a halving of module prices --- and vice versa.

There is probably another 10-fold increase in solar installations in prospect over the next 10 years.  Which will be associated with another halving of the cost of solar.  Meanwhile, the rapid progress of EVS and the need for stationary storage will drive down battery costs, which will continue to halve every four years.

This is irresistible.  The learning curve is being driven by fierce competition, which in turn drives rapid technological advance.  There is nothing Trump or Big Oil or coal miners or the rabid Right can do about this.  They can delay the technological advances in the USA, which will just retard the US economy, but in the rest of the world, the advance of solar plus storage to market dominance in inevitable.  Except in high latitudes.

Dunkelflaute

When I first heard about this phenomenon, I explained it to myself as occurring when there was a cold, cloudy, windless winter's day.  Which isn't too bad a quick descriptor.  But here is a more informative Wikipedia article: 

 

In the renewable energy sector, a dunkelflaute (German: [ˈdʊŋkəlˌflaʊtə] ⓘ, lit. 'dark doldrums' or 'dark wind lull', plural dunkelflauten)[1] is a period of time in which little or no energy can be generated with wind and solar power, because there is neither wind nor sunlight.[2][3][4] In meteorology, this is known as anticyclonic gloom.[5]

Meteorology

Unlike a typical anticyclone, dunkelflauten are associated not with clear skies, but with very dense cloud cover (0.7–0.9), consisting of stratus, stratocumulus, and fog.[6] As of 2022 there is no agreed quantitative definition of dunkelflaute.[7] Li et al. define it as wind and solar both below 20% of capacity during a particular 60-minute period.[8] High albedo of low-level stratocumulus clouds in particular – sometimes the cloud base height is just 400 meters – can reduce solar irradiation by half.[6]

In the north of Europe, dunkelflauten originate from a static high-pressure system that causes an extremely weak wind combined with overcast weather with stratus or stratocumulus clouds.[9] There are 2–10 dunkelflaute events per year.[10] Most of these events occur from October to February; typically 50 to 150 hours per year, a single event usually lasts up to 24 hours.[11]

In Japan, on the other hand, dunkelflauten are seen in summer and winter. The former is caused by stationary fronts in early summer and autumn rainy seasons (called Baiu and Akisame, respectively),[12] while the latter is caused by arrivals of south-coast cyclones.[13]

Renewable energy effects

These periods are a big issue in energy infrastructure if a significant amount of electricity is generated by variable renewable energy (VRE) sources, mainly solar and wind power.[14][1][15] Dunkelflauten can occur simultaneously over a very large region, but are less correlated between geographically distant regions, so multi-national power grid schemes can be helpful.[16] Events that last more than two days over most of Europe happen about once every five years.[17] To ensure power during such periods flexible energy sources may be used, energy may be imported, and demand may be adjusted.[18][19]

For alternative energy sources, countries use fossil fuels (coal, oil and natural gas), hydroelectricity or nuclear power and, less often, energy storage to prevent power outages.[20][21][8][22] Long-term solutions include designing electricity markets to incentivise clean power which is available when needed.[19] A group of countries is following on from Mission Innovation to work together to solve the problem in a clean, low-carbon way by 2030, including looking into carbon capture and storage and the hydrogen economy as possible parts of the solution.[23]

Renewables naysayers say that because of dunkelflauten, we can't use wind and solar to power our economies.  But let's have a look at how often they happen:  a maximum of 150 hours a year, or, about 1.7% of the time.  

Because battery storage is getting so cheap, we soon won't need to use gas peaking plants.  The problem with dunkelflauten is that they last much longer than the 4 to 6 hours when electricity demand exceeds supply, the sort of shortage that can easily be covered by batteries.  For now, until cost-effective long-term storage is invented, we will need to use gas to make sure the lights don't go out.  However, we can make synthetic natural gas from green hydrogen using the Sabatier process, or we may store hydrogen, which we can burn in gas plants for the 150 hours a year when we need it.  These won't add net CO2 to the atmosphere. 

Even if we used fossil gas to provide 100% of the power during dunkelflaute events, and renewables/nuclear the rest of the time, we could still cut emissions from power generation by 98%.  

Because we won't be using dunkelflaute gas plants for most of the year, the cost of their electricity per MWh will be high, because the cost of interest payments and depreciation will be spread over only a few hours of usage.  But, by the same token, taken over the whole year, the occasional high cost per MWh will be spread over thousands of hours of electricity generation.

Dunkelflaute is not an insoluble problem.  We can fix it, and still switch to renewables and cut our emissions.

Pan-Europe wind & solar = stable output

 A most interesting thread from Sarastro on Bluesky.

The past two days we [have] seen something interesting in the European power market: continent wide balancing that is providing security of supply at the lowest prices driven by commercial incentives…

We know that solar and wind and inverse output characteristics. A system that contains both is more secure than one or other alone. This chart from @ember-energy.org makes the point on a European wide scale

 

You can see that on a European wide scale the combined output of wind and solar is less intermittent than solar and wind alone. These charts do not show the risk of hourly balancing though so you still need a source of flexible generation. [Or storage]

This morning we can see that in action. The French grid is importing power from Spain and exporting it to other markets across the French grid in Northern Europe. That’s how you get solar from southern Europe to Northern Europe and wind from the north to the south

 

But take a look at the output of the French nukes: the French have reduced nuclear output in response: they are not just wheeling power across the French system they are managing the French system for cost and using the nukes as a battery

It’s a revelation for those (like me) who have thought of nuclear has inflexible. EDF is showing us that at the heart of the European grid is a huge battery, its nuclear park, capable of firming both south solar and northern wind.

Yesterday we saw something similar with wind from the uk being imported into France and French exports to other European countries

But critically the nukes modulating output…

A couple of points:

1.  I've talked before about how wind and solar tend to balance each other, not just daily, but also seasonally.  It's not perfect, but on a continent-wide grid (as in Europe) the necessary storage/dispatchable power needed (such as gas) is significantly reduced from what would be needed if just wind or just solar was used.
2. Like Sarastro, I also did not know that nuclear could be ramped up and down.  Notice that the percentage moves are small --- roughly 20% --- but because nuclear is so large in European generation, that's enough to go a long way to balancing total grid output.  From the top chart, I estimate the seasonal variability of wind and solar together as ~10% of total output.
3. New nuclear is still much more expensive than new wind+solar combined with 5 hours of storage.  In Australia (without nuclear), 5 hours of storage with 20% overcapacity of wind and solar is enough to provide a stable grid for 99% of the time.    The tricky period seems to occur in July (mid-winter in Australia), when periods of little wind combine with low insolation and high demand for heating, a situation which is called dunkelflaute.  Even though this is a problem only 1% of the time, it would be unacceptable to close down the grid.
4. The solution, until we get better methods of long-term storage, is gas.  Currently, natural gas, but plausibly, in future, synthetic natural gas via the Sabatier process, produced using surplus green electricity.  
5. Alternatively, concentrated solar power (CSP) may do the trick.  Vast Solar, an Australian company, is busy constructing a CSP plant at Port Augusta in South Australia (on the edge of the desert, with lots of sunshine and heat --- CSP doesn't just use light, as solar panels do, it also uses infra-red, otherwise known as heat.)  CSP provides much more storage than batteries (1 hours compared with 4), so is much cheaper for long duration storage.  (Now called Vast Energy, the 30 MW CSP plant is yet to be started, with start-up now planned for Q2/2025.  However, they will now be co-producing green methanol at the plant as well)

Why Europe never has blackouts

 A most interesting analysis.  He shows that without wind and solar, even with maximum demand in mid-winter, the electrical grid in Europe can still cope.  He discusses storage (pumped hydro, with batteries growing fast) and the trans-Europe grid.

He makes two points.  The first is that solar is never zero during daytime, but wind can be zero for a prolonged period.  This means that Europe will still have to "burn things" to make sure it always has enough power.  This implies long-duration storage, if they are not to use gas. He mentions synthetic gas, but doesn't go into detail.  He may mean green hydrogen, or synthetic "natural" gas (green methane) made from green hydrogen via the Sabatier process.   

It seems to me that Europe needs to add more solar from sites in Southern Europe (Spain, Italy, Greece, etc), as solar's winter lows can be compensated for by excess capacity and its nighttime absence by storage.  

Is renewable energy cheaper than fossil fuels?

Answer:  Mostly, yes, but there are complications

From The Climate Brink.

Is renewable energy (RE) cheaper than fossil fuels?

To begin to answer this, we need to define what cost we’re talking about. Let’s first talk about the cost of RE energy vs. fossil-fuel energy on a grid that’s dominated by dispatchable power, such as fossil fuels. This is what most electrical grids are like today.

 

For a grid dominated by dispatchable power (i.e., power sources that can be turned on or off at will), the intermittency of wind and solar imposes no costs. Thus, the relevant cost comparison is between the so-called Levelized Cost of Energy (LCOE) of the various energy sources:

Virtually all credible analyses agree that RE has the lowest LCOE. Therefore, it is the cheapest energy source for grids that contain a lot of dispatchable power.

 

This explains why, for example, 95% of the power scheduled to be hooked up to the ERCOT (Texas) grid is RE (solar, wind, or batteries). Natural gas is 5%.

 

For a grid that’s mainly fossil fuels, every kW of renewable power (RE) you add displaces a kW of expensive and dirty fossil fuel power. But, as the grid gets more and more RE, that changes. At high levels of RE deployment, the intermittency of the wind and solar means that you need to add several kW of wind and solar to displace a single kW of fossil fuels. This drives up the marginal cost of RE energy.

 

In addition, high RE levels mean that RE is competing with the most efficient and cheap fossil-fuel generation, some of which have not yet been paid off. Additionally, the more RE you add, remaining RE sites are higher cost and lower quality.

The net result is that, beyond some point, the price of energy on the grid starts increasing as you add RE. Qualitatively, the price of electricity vs. RE deployment looks like this:

Right now, around 20% of our electricity comes from wind and solar and this is already saving consumers billions of dollars a year. As we increase RE deployment, the price of electricity will continue to decline and consumers save money.

 

Then we reach the minimum price point. One study from NREL concluded that this occurs when RE penetration reaches 57% (in 2050). At this point, electricity produced on this grid is cheaper than a fossil-fuel heavy grid and, as a bonus, we’re also emitting a lot less CO₂.

 

As we move beyond 57%, the declining value of wind and solar to the grid means the price of energy increases. However, it remains below what we’re paying today for a fossil fuel grid until we get to around 90% RE.

 

Let me repeat for those in the backrow: we can get to a 90% RE grid and pay about the same as we’d pay with a fossil-fuel heavy grid. And this doesn’t account for the external costs of fossil fuels (see below).

A significant amount of the discourse about RE focuses on the cost of achieving net zero by 2050, which requires completely eliminating fossil fuels. No one knows how much this will cost, but some studies have produced eye-popping numbers: 

Many analyses have looked at this goal and they agree that a lot of the costs of reaching net zero are driven by the cost of phasing out the last few percent of fossil fuels. The reason is that the last few percent of emissions are the hardest to abate and the ones for which technology to replace fossil fuels is expensive or undeveloped. For example, decarbonizing long-distance airline flights is one of the last things we’ll decarbonize because it would probably require biofuels, which could have very high costs.

 

This is quantified in this plot, which shows the incremental abatement cost (orange line) as a function of how much RE is on the grid. For a 95%-RE grid, the abatement cost is $200/tonne, increasing to $930/tonne for 100% RE.

Figure 1 of Cole et al.

Thus, it’s easy to look at the price tag for getting to net zero and conclude, “Wow, this is too expensive.” But that misses the fact that the cost of getting to a slightly lower value, e.g., a 90% clean-energy grid, is actually quite modest.

 

These net-zero estimates also hinge heavily on future innovation — a variable notoriously difficult to predict. History has shown us, particularly in the last decade, that technological advancements can drastically outpace predictions, as seen with the enormous drop in the cost of solar panels, which no one predicted.

 

External costs

All of these discussions focus on the market price of energy. Such a discussion neglects the extensive subsidies that distort the energy market. While RE sources receive financial support, the subsidies for fossil fuels are substantially larger and more ingrained within global economies.

 

Moreover, the price of fossil fuels seldom reflects their full societal costs — what economists call externalities. Recent estimates of the cost of climate impacts puts it around $185/ton of CO₂ emitted. These costs are not included in the cost of fossil fuels.

 

Air pollution from fossil fuels kills millions of people every year. Like the climate impacts, the costs of this are not included in the price. Fossil fuels have also been linked to significant political and social instability. For instance, the U.S. invaded the Middle East twice in the last 35 years in order to stabilize the oil supply. The Russian invasion of Ukraine is intimately tied to fossil fuels. These costs are also not included in the costs of fossil fuels.

 

If we added these externalized costs to the cost of fossil fuels, the argument increasingly tilts in favor of RE.

 

Summary

In any complex discussion, you need to carefully define the question you’re asking. Much of the discussion around renewable energy focuses on net zero, because that’s what we ultimately need to aim for. We really don’t know how much achieving net zero by 2050 will cost because it will depend to a large extent on future innovation.

 

But a large chunk of the cost of net zero is driven by the last few percent of decarbonization. If you talk about, say, a 90% clean grid, the cost of achieving that using today’s technology is approximately zero. And this cost comparison excludes the external costs of fossil fuels: climate impacts, air pollution, geopolitical instability. Taking all factors of those into account, there’s no question that we can largely decarbonize today and end up with a better economy and cleaner environment.

 

I found this article extremely enlightening.  It put into words something which I had only intuitively understood.  But it highlights how the cheapest RE now might not be the cheapest RE in 2040 (ignoring of course technological advances).  For example, CSP (concentrated solar power) is more expensive than solar panels.  However, it can produce dispatchable electricity, which means it will be very valuable for that last 10% of electricity supply.  Another example:  SMRs (small modular reactors) are prolly 3 or 4 times as expensive as solar, now.  But to convert solar into dispatchable power would require at least 12 hours of storage.  That's still very expensive, though no doubt battery costs will continue to decline.  But so might the costs of SMRs.  Again, for that last 10%, SMRs might be the answer.    Yet another example: power-to-gas.  Using green methane (produced using green electricity to make hydrogen by electrolysis, which is then converted into methane by the Sabatier process) to run peaking gas plants will be expensive, but again, makes sense for the last 10%.

We may have an answer sooner than we thought.   The state of South Australia has been steadily increasing the percentage of renewables in its grid for 17 years.  So far this year, it has averaged 75%.  It could reach 100% within 5 years.  Now, this isn't a perfect test for how high renewables can go, because SA (unlike, say, Texas) has high-voltage links with other grids, so it can buy or sell power to the other states.  Nevertheless, we will get a clear idea of the issues quite soon.