A provocative title, but a question that deserves more consideration than it receives – for reasons I will outline.
By “renewables”, I specifically mean solar power and wind.
The thesis in short:
A grid with lots of renewables is a grid that is reliant on natural gas.
In principle, renewables could be combined with storage, but this is much more expensive than widely understood.
I see a divergence between countries that pursue a) renewables + storage (more realistically, this is renewables + gas) and b) nuclear power.
Group a) is vulnerable politically, both externally – due to reliance on natural gas, and internally – due to popular unrest with volatile energy costs. These countries/regions also suffer economically because higher energy prices are bad for business. Group b benefits from the opposite effects.
Much like with vaccines, a population that applies science and engineering principles to develop safe technology will be at a strong advantage.
Lastly, technology has always developed towards higher density, i.e. packing more information and energy into smaller spaces (think circuit boards and mobile phones and the discovery of oil/gas and engines). Wind and solar are the opposite of this; they harness low density energy sources. Thermodynamically, the highest efficiency and lowest cost systems are found and developed with highest energy densities.
I finish this piece with some reasons as to why this thesis may be wrong.
A grid with lots of renewables is a grid that is reliant on natural gas
As of March 2022, Germany cannot afford to stop buying Russian gas. The grid would shut down.
Wind and solar provide intermittent power, so an alternate source of power is required when wind and solar are unavailable.
Neither nuclear power nor coal are ideal complements to renewable energy.
Both nuclear and coal power plants are slow to ramp up and down. In simplified terms, it takes those processes a long time to get warmed up and cool down. If the wind falls, they are too slow to ramp up production.
Gas is an ideal complement because it can be turned on quickly
Open cycle gas power plants can be ramped up quickly. Quite simply, the gas is burned inside a turbine, and the turbine turns to create power. This is unlike coal and nuclear power, both of which generate heat that is then used to heat water to steam, and that steam is used to drive a turbine.
Open cycle gas power plants are also cheap to build compared to coal or nuclear plants. This is important when complementing wind and solar power because the gas plants only run part of the time. They earn money less of the time, so they need to be cheaper to start with.
Anything that complements wind and solar – whether it’s gas plants or storage – needs to be cheap in capital cost, because it will only operate part of the time. This is the capacity factor problem with renewables and any technology the complements renewables – each much operate highly intermittently and amortise capital costs over less output.
In principle, renewables could be combined with storage, but this is much more expensive than widely understood
The capacity factor problem
When starting with a grid that has natural gas, it looks cheap to add in small (or even moderate) amounts of renewables. You turn off the gas plants when the wind is blowing or the sun is shining.
If you approach the problem from the other direction – starting from scratch with a grid that has no gas – the costs of intermittency become clear.
If one builds a grid from scratch with nuclear or coal power (I’m not a fan of coal due to pollution and also carbon emissions), one can size the power plants to meet the maximum demand of the grid during the day.
If one builds solar or wind to meet the grid’s demand, there’s a need to build not only enough to meet the peak demand, but also generate enough power to store at other times. So, the installed base (measured in Giga Watts) of power in a fully renewables grid is multiples of what it would be in a grid powered by a power stable source like nuclear or coal.
Moreover, when installing more solar (or more wind) in the same region, the power produced by new units of solar (or wind) power increasingly tends to overlap with what is already installed. Increasingly, there are periods where turbines or panels need to be curtailed in the power they provide. This hurts project economics and the cost of electricity.
In short, by adding renewables to a grid powered by gas, the gas is heavily subsidising the variability. If instead one comes from the direction of starting with a blank slate, the cost of renewables is seen to be much greater because renewables and, consequently, any technology that is used to balance the variability of renewables have a low capacity factor (they each only operate for part of the time).
This brings us to the role of storage.
The storage problem
A grid powered by solar and wind requires storage on multiple time-scales. There is a need for storage from time scales all of the way from minutes, to months. These longer time scales, in particular, pose an economic problem for storage.
The economic problem of long duration storage is demonstrated through a simple example:
Let’s say you have 1 kWh of storage to offer, and you offer it out at €0.15 (about the price of electricity at home in Ireland) every time someone charges.
If your battery is used (i.e. charged and discharged) once per year, you’ll make €0.15 from your battery every year.
If your battery is used (i.e. charged and discharged) once per hour, you’ll make about €1,300 from your battery every year (at which point you probably replace the battery because it’s close to 10,000 cycles).
So, for long duration storage (the once per year example) the battery earns very little money. For short term storage, the battery earns lots of money. Flipped around, in owning a battery (or any form of storage, for that matter), one is happy to offer frequent (i.e. short term) charging at a much lower price than long term (infrequent) charging.
To illustrate this further, if one takes a price of €100 per kWh of storage for batteries today and assumes an operating life of 10 years (both optimistic), one can generate a cost curve for storage per kWh of charge/discharge delivered:
As per the above chart, hourly and daily storage can make sense (i.e. the cost per kWh is small or comparable to the average price of electricity). However, once storage duration gets up to weekly or monthly storage, it’s clear that doing storage at all results in the cost of electricity delivered being doubled or much worse.
Of course, there is the argument that: “Maybe this is the price we have to pay for energy security and eliminating carbon emissions?” I think that’s a fair question and one that should be considered. However, there is the counterbalance that it is strongly in our interests to keep energy prices low for two reasons:
The general public is highly sensitive to high, and especially volatile, energy prices. This empirically true if one looks at the history of fuel subsidies and protests.
High energy prices hurt domestic competitiveness and lead to the off-shoring of energy intensive businesses.
Side note: Some people worry about having too many data-centers sucking energy. I worry about having too few. Data-centers mean more software in the economy, and more software generally means less hardware and less energy overall for the economy. More software also means more technological progress.
A brief note on hydrogen
To transition from fossil fuels, we may well need to move to hydrogen or ammonia or some other high density fuel for certain applications. I think it’s possible that we see 10X or 100X improvements in batteries, so I don’t rule those out either – even for airplanes. I also don’t rule out nuclear on a micro-scale – in fact this seems the most likely winner to me on a 100 year time horizon.
What I don’t see with hydrogen today, is a solution to the variability of solar and wind. What is widely misunderstood about hydrogen (and many other storage technologies) is that it requires capital cost to build hydrogen production facilities. If one has 3 GW of wind, and 2 GW is to be stored, that requires 2 GW of hydrogen generation.
Consider a baseline of having 1 GW of nuclear power, and now consider replacing that with wind and hydrogen.
First off, probably 3 GW (and likely more) of wind is required because you need to have enough to store energy for when there is no wind.
In addition, there needs to be 2 GW of power dedicated to hydrogen production. Unfortunately, hydrogen can only be produced when the wind is blowing, so – most of the time – all of that hydrogen production equipment (probably an electrolyser) is sitting idle. That time idle means the hydrogen equipment isn’t earning an economic return, and that drives up the cost of doing the hydrogen storage.
If the hydrogen is to be used as storage for the grid energy (fewer people suggest this because they rightfully think about hydrogen as directed to storage), then there also needs to be at least 1 GW worth of hydrogen turbines to turn the hydrogen back into electricity.
So, we have:
1 GW of nuclear capacity
3 GW of wind capacity + 2 GW of hydrogen production + 1 GW of hydrogen turbine + other hydrogen liquification + de-liquification + storage, and I think that is optimistic.
One can and should debate the costs per GW of the capital and then consider operational costs, like waste management for nuclear. I’m deferring that discussion until later in this piece.
In short, I’m a fan of hydrogen or ammonia or similar for future fuels, but don’t think they solve the energy storage problem with renewables.
Much like with vaccines as with nuclear power, a population that applies science and engineering principles to develop safe technology will be at a strong advantage.
The three core issues raised with nuclear power are:
Danger of accidents (e.g. Chernobyl)
How to deal with nuclear waste
Risk of nuclear weapons
Sourcing nuclear fuel
I would happily live by a well designed nuclear power plant. Many people do. I did for a short time in France – we have family friends where the guy worked his whole career at a nuclear power plant.
Not all nuclear plants or technologies are the same – some are substantially better than others. If the US or EU had applied the same science and engineering process to vaccine development as was applied to building Chernobyl, I’m confident there would have been massive deaths.
I believe that safe nuclear power plants are within humanity’s science and engineering capabilities and I would happily live near a nuclear plant such as those in France, Germany, the UK and many other countries.
Dealing with Nuclear Waste
Gas turbines (combined cycle type) can achieve 40% or even 50% thermodynamic efficiency.
The efficiency with which we convert nuclear fuel to power is much smaller than that. Still, nuclear power plants require about a truck load of fuel every year or so. My point, though, is that there is room for dramatic improvement in how we convert nuclear fuel to power – and dramatic room for improvement in reducing what is generated as waste.
I would go so far as to say it is perhaps possible that the majority of nuclear waste has already been produced in the past – even if we were to ramp up nuclear production. Our efficiency in converting nuclear fuel to power is increasing rapidly, as is the rate at which we can reduce waste.
The reality is, that we already have generated significant nuclear waste. I don’t believe that expanding nuclear will substantially increase the total waste generated – certainly not by orders of magnitude.
To the contrary, perhaps the worst thing we could do for handling historical nuclear waste is to stop our technical development of nuclear power.
Risk of Nuclear Weapons (and of badly built plants like Chernobyl)
It seems naive to think that stopping the development of nuclear power will stop adversaries from developing nuclear weapons or badly building/running their energy infrastructure.
Adversaries will always be there. Technological regression does not seem a winning or even sustainable strategy.
Sourcing nuclear fuel
A core piece of my thesis is that relying on gas leaves a country reliant on other countries (unless they have their own gas supply).
The supply of nuclear fuel is also concentrated, and can lead to geopolitical dependence. However:
It’s easier to store 18 months worth of nuclear fuel than of gas. So, nuclear is much more sheltered to price volatility than gas.
Nuclear has more room for technological improvement than gas, so there is more room to reduce reliance on sources of fuel.
Reusing nuclear waste as fuel can provide a hedge against foreign fuel sources.
Technology has always developed towards higher density
Every year, we fit more onto semiconductor chips. Our phones get smaller and more powerful. We pack more information (or “order” as a thermodynamicist would think about it) into smaller spaces.
This was the same with the discovery of fuel and the automobile. Fuel is incredibly high in energy density. This energy density allows us to build useful things without taking up too much space or requiring too much energy. Imagine what it would have been like to build a car if petrol were 100 times less density. Imagine the cost of the fuel storage tanks, imagine the cost of the engine.
Thermodynamically, high energy densities (analogous, and equivalent thermodynamically in many ways to information density) allow us to do more with less.
Renewables in this sense are the opposite of energy density. Neither the sun nor the wind are particularly energy dense. Each require significant areas of panels or turbines to extract energy.
To this, one might argue that it’s ok to be low energy density if the materials can be cheap. I think this is true – in the short term – and then we revert to the earlier cost discussions I laid out above.
However, in the long term – allowing for technology – what matters is the potentially for 10X or 100X improvements. Generally, I think 10X or 100X improvements are much more likely to come from going to higher density rather than lower density energy sources. Big improvements are more likely in nuclear than in solar or wind.
Of course, it’s possible we may see a material much cheaper than silicon for solar, or much cheaper and energy dense than lithium ion batteries, or much lighter in weight and cheaper and stronger than metals for wind turbines.
However, the costs of a fully wind/solar grid with storage seem much higher than nuclear today, and the costs of wind/solar/gas much more volatile and politically vulnerable. This, coupled with room for improvement, makes the nuclear picture more compelling for me.
I thought a lot about nuclear, particularly for Ireland – a country that is small, without nuclear experience and with a historical aversion to nuclear. However, even for a country as small as Ireland, I now think that considering (but not going all in on) a long term strategy for nuclear power is sensible. It seems smart to look at today’s realities, and couple them with a progressive view of technology in the medium to long term future.
A divergence between countries that pursue A) renewables and B) nuclear power.
Ireland, the UK and Germany – as examples – are countries that think (maybe thought?) they are on track A for renewables plus storage, but are really on track a for an unstable future of renewables plus gas.
France and Finland are countries on track B and this will bring them relative political security and econonic strength. (Not a coincidence that Finland currently buys lots of gas from Russia and is not a member of NATO.)
Where might I be wrong?
Broadly, I could be wrong technically or I could be wrong politically. Humanity has progressed so rapidly that I think politics/culture are downstream of technology, so being wrong technologically is probably the bigger mistake over a 100+ year time period.
High energy prices could be politically tolerable
It’s possible that energy at €0.40 – €1 (as opposed to €0.17/kWh at my house today) is tolerable. Perhaps there are ways to compensate those who are hardest hit by energy prices. This is not a new problem and it doesn’t seem great solutions have been reached historically other than government fuel subsidies (or installing nuclear power).
Maybe certain countries can operate quite well with high energy prices and industries that don’t require much energy – outsourcing high energy industries elsewhere. To a large degree, this is what has happened with polluting industries – they have been outsourced to poorer countries. Ironically, in a country that uses a low amount of energy, there is less economy of scale on energy production, so arguably the price per unit of energy could be higher if one incentivises companies/industry that require less energy.
There is certainly some argument that energy security and/or reducing carbon emissions are critical to the extent that we should just splurge and solve the problem. My sense is that the challenge is not just the economic cost, but the coordination problem of implementing such a large scale project. All technological approaches are likely hard, but all are not equal in terms of coordination issues. Building a small number of nuclear plants seems more tractable to me than a wide network of solar + wind + batteries.
Underestimating technological progress
It has often been a mistake to underestimate technological progress and overestimate resource scarcity. I may be making this mistake here too on batteries, solar, wind and/or other forms of storage.
Perhaps the most likely technological reason I’m wrong is if batteries drop in price by a factor of ten within ten years. At that point, longer term storage becomes more feasible. Still, the roll-out of such a large amount of batteries and wind/solar farms is a vast project in terms of land-area, and not one that is easy to do cost wise or from a planning permission perspective.
Am I missing other options?
Common suggestions might involve:
Demand management – but I don’t think most grid operators think changing consumer behaviour is realistically going to balance wind/solar
Advanced geothermal – This involves digging deep into the earth’s core where temperatures are very hot, and using that to make steam to drive turbines. This strikes me as high density and also a steady source of power. It seems worth investigation. It’s not as dense as nuclear, so I still favour nuclear long term.
Nuclear fusion – I include this under the broad “nuclear” umbrella as what I think is most likely the best long term solution.
Reduce energy consumption – Reducing consumption doesn’t solve the issue of intermittent power supplies. I believe technology has and will dramatically increase energy efficiency, but we should be careful about curtailing energy use. Paradoxically, curtailing energy use can curtail progress and technology that allows sustainable growth.
Grid interconnection – this does reduce variability, but interconnection – by definition – increases dependence and doesn’t increase security. Interconnection also comes at a cost and often with planning permission. Generally, I think doing as much as possible makes sense, but there are natural constraints around rates of deployment and distances over which it makes economic sense that limit the ability of interconnection to fully address renewables’ variability.
Other forms of storage (e.g. compressed air or thermal storage). Generally, I see many of these forms of storage as being low energy density, difficult to improve by a factor of 10X, and inferior to technological progress in a high energy density direction.
Battery storage in electrical cars. A 50 kWh car battery is quite a lot of energy storage. If every household in Ireland (say, 1 million) had one, that would be 50 GWh of storage. That could provide an average load of about 2 GW to the Irish grid, which turns out to be on the same scale as the Irish grid. Being able to harness this storage would require significant grid upgrades (beyond my understanding), but seems theoretically possible. Perhaps this is a blind spot for me that supports renewable energy – although it doesn’t address the cost of added wind and solar capacity required to charge this storage.
Carbon capture – coupling this with a long term domestic source of gas could provide some stability and perhaps should be considered. I don’t think gas is the most economic long term option. I think it will be beaten by nuclear.
The war in Ukraine will (or at least should) make countries going down the pure renewables track think more seriously about the reality of building energy storage to replace the role of natural gas. This has already led to some reconsideration of nuclear power.
I support renewables and batteries, but only at low grid penetration, perhaps less than 20%. I’m hesitant to support renewables at a higher level because renewables make the grid dependent on gas and I think energy storage is unrealistic today – economically and practically.
Insufficient attention is being given to stable supplies of power, with nuclear as the obvious option to consider. Countries that go continue or start down the path of engineering safe nuclear plants will be at a big advantage.
Questions very welcome to ronan at [this domain], and I’ll aim to make any corrections as they inevitably are required.