How Much Steel Will All Those Wind Turbines & Solar Panels Need, & Can We Make It?

In recent days I’ve finally made my first scenario projection for steel demand, steel technologies which will dominate, and decarbonization curves as a result. All through 2100, as is my wont, with explanations about what forces will bend the curves. As with everything, it’s a thought exercise and a way to think about the future so that investors, policy makers, and corporations can make better decisions today.

In that process, I’ve stumbled across some interesting data points. One of them was that there are about 350 million tons of steel embodied in US pipelines, four times the annual demand for that major economy. I expect that a lot of that good quality steel will be extracted and turned into useful things like wind turbines. Of course, commenters pointed out that the oil and gas industry coated a lot of the pipelines built between 1920 and 1980 with asbestos, and in a lot of places they are buried so it’s questionable about whether it would be worth digging them up. Another commenter pointed out EPA material indicating that the interior of many was scaled with sludge containing things like uranium, thorium, and radium and their decay products, potassium-40 and lead-210/polonium-210.

It appears that some pipelines are better thought of as toxic waste, and in many cases will be treated as such instead of treated as a scrap asset. As my calculations were intended to be evocative, not prescriptive, I’m not particularly worried about it, but am now a bit more worried about the toxic stew that’s going to be left to rust in the ground.

The evocative point is that we’ve manufactured an awful lot of steel and will be manufacturing a lot in future years, but that increasingly we’ll be scrapping steel products we’ve already made to supply steel globally. China is coming to the end of its massive infrastructure urban infrastructure build-out, a major demand source for steel, is deeply unlikely to ever get to western levels of car ownership, and will be transitioning to a steel scrapping and recycling economy much more. Parts of the world that are still developing won’t be doing it as quickly as China did, nor do they have approaching zero infrastructure as China did in 1978 or 1.4 billion people. So in my projection, the manufacturing and demand curve will flatten and be much more by scrap steel running through electric steel minimills powered by renewable electricity, with obvious climate benefits.

But in reviewing material from ArcelorMittal, a $80 billion annual revenue steel multinational headquartered in Luxembourg but owned by Indian firm Mittal Steel after a 2006 takeover, I found another fun factoid.

Each new MW of solar power requires between 35 to 45 tons of steel, and each new MW of wind power requires *120 to 180 tons of steel. *Applies only to steel in offshore wind foundations.

That was sufficiently interesting to trigger me to ask how much that turns into to meet all steel demand for all energy needs for the world, assuming wind and solar only. Once again, a thought exercise intended to be evocative, not an assertion that they will provide more than the large majority of energy.

The numbers don’t include onshore wind, and the requirements there are lower as the foundations are at ground level. Solar is likely to end up providing a bit more of the energy than wind due to its absurd simplicity and numerical scaling advantages. As noted, wind and solar both have absurd scaling and modularity advantages compared to nuclear and other generation technologies, so it’s not like there will be a lot of other new generation in the end mix.

The combination led me to pick 70 tons of steel per MW as the average in the future.

The next question is how much energy is required? Recently, I published one of my more read assessments, that of the value of electrification of energy in the US with heat pumps, some additional industrial heat electrification wedges, and transportation. Of course, this includes getting all US energy needs in the form of electricity from renewables for the most part, with the only rejected energy being transmission and storage losses. In that scenario and holding everything else — population, GDP, comfort — the same, US primary energy requirements would drop by 50%.

Applying that coarse napkin take globally, total primary energy consumption annually today is in the range of 595 exajoules, suggesting that we would need perhaps 300 exajoules in a fully electrified global economy. That turns into about 84,000 TWh per year, which is a lot, but for context China generates about 8,000 TWh annually right now, and probably about 2,000 TWh of that is from renewables already. When we get rid of deeply inefficient fossil fuels and use electricity for everything possible, energy requirements plummet.

Next I took a blended capacity factor of 30% for the mix of solar and onshore and offshore wind energy. That means we would need about 32 TW of wind and solar deployment.

At 70 tons of steel per MW, that turns into about 2,200 million tons, which seems like a lot.

Millions of Tons of Steel Per Year By Method Through 2100, chart by author

However, let’s contextualize 2,200 million tons. Total steel production globally in 2020 was 1,878 million tons. Less than 1.2 years global steel manufacturing at today’s volumes would enable manufacturing of all wind and solar for all of global energy requirements.

My projection for 2030 is about 2,000 million tons per year, aligning with continued but no longer accelerating demand from China for another few years and increasing demand from India, the number two manufacturer and consumer at present. My projection is that peak steel production and demand will come in 2060 at 2,300 million tons of steel, so at peak steel only a single year’s output would manufacturer all of the wind turbines and solar panels required for an electrified, decarbonized society.

2050 is 27 years away and we’re already building 270 GW of new solar and approaching a GW of wind nameplate capacity annually. That’s been increasing exponentially, and obviously needs to increase a lot more, but 1.2 TW a year is about 80 million tons a year of steel demand, well under current manufacturing. In fact, it’s in the range of the global automotive and truck industry which saw demand of around 70 million tons in 2020, and in my projection for 2030 has 100 million tons demand.

Obviously we are going to massively increase renewables deployment in an s-curve through 2080 or so, so the steep part of the curve from 2030 to 2050 will see more than 80 million tons demand, but still falls well within my total steel demand projections through 2100.

And as I keep pointing out, one of the world’s largest consumers of steel is the fossil fuel industry, between the aforementioned pipelines as well as drill rigs, processing facilities, refineries, trucks, buildings, marine oil tankers, and train oil cars. That demand is going to slope downward in the coming decades as demand from renewables ramps up. What percentage of steel exactly does the fossil fuel industry consume annually? It’s difficult to say, as public statistics I’ve been able to find don’t include industry consumption breakouts. If any reader has good data, please share it.

Finally, back to those pesky pipelines again. Just US pipelines, with the caveats mentioned, have 350 million tons of steel embodied in them. We only need six times as much as that to build all the wind and solar we need to decarbonize all energy for the world.

Steel will not remotely be a constraint for global transformation of energy over the coming decades. We make vastly more of it per year than annual requirements, and there’s an increasing amount of scrap being created.