How should we generate electricity?
Currently, the majority of our electricity comes from fossil fuels, especially coal and natural gas. Burning fossil fuels has given us access to tremendous amounts of energy and has made modern civilization possible. Without them, we would have had trouble feeding ourselves, let alone obtaining our current standard of living.
Unfortunately, burning fossil fuels releases greenhouse gases that warm the global climate. And they will run out eventually. What other sources are available?
Prerequisites: Some knowledge of what the different power sources are might be helpful.
Originally Written: March 2021.
Confidence Level: Numbers are approximate, but are the right order of magnitude. Finding all of the sources for this was as hard as writing it, so feel free to check them.
Some Background Information
Power is measured in watts, kilowatts (1 kW = 1,000 W), megawatts (1 MW = 1,000,000 W), and gigawatts (1 GW = 1,000,000,000 W). Power is energy divided by time. To get back to energy, you multiply by time. While energy is often measured in Joules (1 J = 1 W $\times$ 1 sec), electrical energy is measured in kilowatt hours (1 kW-hr = 3,600,000 J).
Electricity is only one way we use power. We also use natural gas to directly heat our homes, burn coal in factories, and use oil in our cars. Most other power could use electricity if we wanted them to. I will only focus on electricity. The other sources are also important, but I am less familiar with them and this will be long enough as it is.
I will be presenting a simple model to show how various strategies to move away from fossil fuels could work. The simplifications will make it look easier to move away from fossil fuels than it actually is, but I try to be fair between the different alternative energy sources. Technological improvement is possible for any energy source, especially if we start building more of them, so I will not try to guess how fast the technology will advance. All costs will be based on current (2019-2020) technology.
To give you some intuition, here is the amount of electrical power produced by several countries and states:
USA | 489 GW | Germany | 65 GW | China | 888 GW | ||
Texas | 54 GW | France | 60 GW | India | 178 GW | ||
California | 22 GW | UK | 36 GW | Japan | 115 GW |
Alternative Power Sources
Hydroelectricity
The first electrical grids in New York and Grenoble were powered by hydroelectricity. Hydroelectricity is still great. It does not produce any greenhouse gases and we can control how much electricity is produced, except for during droughts and floods.
In the developed world, most of the hydroelectric capacity has already been built. Hydroelectricity current supplies about 10% of our electricity. There are some improvements we could make, but it will never be a majority of our power supply. It is extremely important locally, in areas with lots of rain and mountains, but does not produce power on a continental or global scale. New hydroelectric capacity in the developing world is more promising both because fewer dams have already been built and because electricity use is currently lower.
Wind
Wind is also an old power source. Unlike hydroelectricity, which has always been important, wind was partially forgotten about and recently rediscovered. The cost of wind has fallen by 40% from what it was 10 years ago.[2]This is for onshore wind. The cost of offshore wind has “only” fallen by 29%. Source.
Building 1 MW of wind capacity (about 1 large turbine) costs on average $1,800,000,[3]This is the base overnight cost for wind (onshore). Source. requires about 75 acres of land,[4]The total project area of a wind farm is 34 $\pm$ 22 hectares (84 $\pm$ 54 acres), of which only 0.3 $\pm$ 0.3 hectares (0.8 $\pm$ 0.8 acres) are covered by the turbines and service roads themselves. … Continue reading and lasts for about 20 years.[5]Source. Wind turbines cannot be put too close together or they will interfere with each other, but the land under them can be used for other purposes.
Turbines only produce electricity when the wind is blowing. On average, the wind is stronger at night and during the winter.
Solar
Solar technology has advanced even more quickly than wind. The cost of solar panels has fallen by 82% from what is was 10 years ago.[6]This is for photovoltaics. The cost of concentrating solar power has “only” fallen by 47%. Source (same as $\uparrow 1$). Solar is significantly more competitive now that it costs only a fifth as much.
Building 1 MW of solar capacity costs on average $1,200,000,[7]This is the base overnight cost for solar photovoltaic with tracking. Source (same as $\uparrow 2$). requires about 30 acres of land,[8]Various solar technologies range from 25-50 acres for 1 MW. Source. and lasts for about 25 years.[9]Manufacturers’ warranties are for 25 years. Since this is a gradual degradation at about 1% per year instead of a sudden failure, they can be used beyond this. Source.
Solar panels only produce electricity during the day. They produce less electricity in the morning and evening and when it is cloudy.
Nuclear
Nuclear power does not produce greenhouse gases and does not contribute to climate change. While nuclear fuel is not renewable, it has extremely high energy density and we have enough of it to last for millions of years.
The cost of the fuel itself is negligible. The cost of construction is 75% of the total cost over the lifetime of the power plant. For comparison, construction costs are 20-65% of the cost of fossil fuels.[10]Fossil fuel plants cost less to build than nuclear reactors and fossil fuels are more expensive than nuclear fuels for the amount of energy they provide. Source. I will only focus on the construction costs, not the maintenance and fuel costs, for nuclear, solar, and wind.
Building 1 MW of nuclear capacity costs on average $6,000,000.[11]This is the base overnight cost for a light water reactor. Source (same as $\uparrow 2$). The amount of land needed is similar to fossil fuel power powers. Nuclear power plants are designed to last for 40 years, but can last longer.[12]The average age of US nuclear power plants is almost 40. Of the 93 nuclear reactors in America, 88 have been approved to last for 60 years and 6 have been approved to last for 80 years. Source. Each nuclear power plant produces about 1 GW of power, so 500 of them could produce the average electrical power used by the United States.
Nuclear has one additional problem: it is scary. The risks of nuclear power are extremely unlikely, but potentially terrible. To keep this post from getting too long, I will discuss these concerns in a separate post. If you systematically compare how many people have died and how much electricity has been produced, nuclear is the safest power source.[13]Globally, nuclear is about 1,000 times safer than coal and even safer than wind or solar. Source.[14]EDIT [2022-02-06]: Our World in Data claims that solar is the safest energy source. Both agree that nuclear and renewables are orders of magnitude safer than fossil fuels or biomass. Source.
Fusion
Fusion would be an even better power source. It does not produce greenhouse gases. Its fuel has even higher energy density than nuclear fuel and is even more abundant. There is no proliferation risk. The only problem is that we haven’t figured out how to do it yet.
I predict that we will get fusion within the next 15 years (80% confidence) to 20 years (90% confidence).[15]For “get fusion”, I mean $Q>5$ for a magnetically confined plasma, which should be enough to build a power plant with $Q_{eng}>1$. These seem like bold predictions to many people, … Continue reading The first fusion power sold to the grid will likely occur within 5 years after that. Optimizing fusion power plants and making them less expensive will probably take another decade or two (pure speculation). Once fusion becomes feasible, we should use it as our primary source of electricity.
For more details, I recommend The Future of Fusion Energy by Jason Parisi and Justin Ball (2019).[16]Book review coming.
Others
Other alternative energy sources are sometimes mentioned, like tidal or wave or geothermal[17]I have recently heard good things about geothermal, so I probably shouldn’t dismiss it so quickly. electricity. These energy sources are too small and diffuse to be used on a continental or global scale. They can be effectively used in a few locations, like near volcanoes or on sparsely populated islands, but they will not supply a majority of our electricity.
Biofuels could, in principle, provide a majority of our electricity, but they have a fundamental problem. It is always more efficient to cover a field in solar panels than to grow biofuels. Plants use sunlight for things other than producing electricity. Even on the best farmland, biofuels are only about half as efficient as solar panels. If you wouldn’t cover a farm with solar panels, you should not use it to grow biofuels.
While there are many power sources which can be important to use locally, the only options currently available for continental or global electricity production are fossil fuels, nuclear, solar, and wind.
The next post will look at power sources from the perspective of the grid.
References
↑1 | Source for countries. Source for states. Note that I converted Gigawatt-hours / year to Gigawatts. |
---|---|
↑2 | This is for onshore wind. The cost of offshore wind has “only” fallen by 29%. Source. |
↑3 | This is the base overnight cost for wind (onshore). Source. |
↑4 | The total project area of a wind farm is 34 $\pm$ 22 hectares (84 $\pm$ 54 acres), of which only 0.3 $\pm$ 0.3 hectares (0.8 $\pm$ 0.8 acres) are covered by the turbines and service roads themselves. Source. |
↑5 | Source. |
↑6 | This is for photovoltaics. The cost of concentrating solar power has “only” fallen by 47%. Source (same as $\uparrow 1$). |
↑7 | This is the base overnight cost for solar photovoltaic with tracking. Source (same as $\uparrow 2$). |
↑8 | Various solar technologies range from 25-50 acres for 1 MW. Source. |
↑9 | Manufacturers’ warranties are for 25 years. Since this is a gradual degradation at about 1% per year instead of a sudden failure, they can be used beyond this. Source. |
↑10 | Fossil fuel plants cost less to build than nuclear reactors and fossil fuels are more expensive than nuclear fuels for the amount of energy they provide. Source. |
↑11 | This is the base overnight cost for a light water reactor. Source (same as $\uparrow 2$). |
↑12 | The average age of US nuclear power plants is almost 40. Of the 93 nuclear reactors in America, 88 have been approved to last for 60 years and 6 have been approved to last for 80 years. Source. |
↑13 | Globally, nuclear is about 1,000 times safer than coal and even safer than wind or solar. Source. |
↑14 | EDIT [2022-02-06]: Our World in Data claims that solar is the safest energy source. Both agree that nuclear and renewables are orders of magnitude safer than fossil fuels or biomass. Source. |
↑15 | For “get fusion”, I mean $Q>5$ for a magnetically confined plasma, which should be enough to build a power plant with $Q_{eng}>1$. These seem like bold predictions to many people, so I will justify them in a future post. The main argument is that there are now multiple plausible independent paths to fusion. |
↑16 | Book review coming. |
↑17 | I have recently heard good things about geothermal, so I probably shouldn’t dismiss it so quickly. |
A physics professor was joking that fusion has been 50 years away for the last 50 years… but now we’re confident it really is only 50 years away!
In your article you say it’ll hit the grid in 20-25 years. Do you think it’ll be efficient and scaleable at that point?
As of about 2017, I would respond to that joke with: “It was 50 years away and always will be. Now it’s 20 years away and already will be. Progress!” I would have put about a 60% chance of fusion (Q>5) by 2035 and 70% by 2040. At the time, “getting fusion” was almost synonymous with “ITER works”.
Things have changed since then, which has dramatically increased my estimate. The most important recent event is the founding of Commonwealth Fusion in 2018, which led to other series startups. This, and the more serious funding of fusion in East Asia, mean that there is no longer only one plausible path to fusion. We can afford some failures.
An individual fusion power plant would have to be big, so scalable isn’t that big of a problem. I don’t know how economical they will be then.